biological and medical physics, · biological and medical physics, ... they lie at the crossroads...

biological and medical physics,biomedical engineering

biological and medical physics,biomedical engineeringThe fields of biological and medical physics and biomedical engineering are broad, multidisciplinary anddynamic. They lie at the crossroads of frontier research in physics, biology, chemistry, and medicine. TheBiological and Medical Physics, Biomedical Engineering Series is intended to be comprehensive, covering abroad range of topics important to the study of the physical, chemical and biological sciences. Its goal is toprovide scientists and engineers with textbooks, monographs, and reference works to address the growingneed for information.

Books in the series emphasize established and emergent areas of science including molecular, membrane,and mathematical biophysics; photosynthetic energy harvesting and conversion; information processing;physical principles of genetics; sensory communications; automata networks, neural networks, and cellu-lar automata. Equally important will be coverage of applied aspects of biological and medical physics andbiomedical engineering such as molecular electronic components and devices, biosensors, medicine, imag-ing, physical principles of renewable energy production, advanced prostheses, and environmental control andengineering.

Editor-in-Chief:Elias Greenbaum, Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA

Editorial Board:Masuo Aizawa, Department of Bioengineering,Tokyo Institute of Technology, Yokohama, Japan

Olaf S. Andersen, Department of Physiology,Biophysics & Molecular Medicine,Cornell University, New York, USA

Robert H. Austin, Department of Physics,Princeton University, Princeton, New Jersey, USA

James Barber, Department of Biochemistry,Imperial College of Science, Technologyand Medicine, London, England

Howard C. Berg, Department of Molecularand Cellular Biology, Harvard University,Cambridge, Massachusetts, USA

Victor Bloomfield, Department of Biochemistry,University of Minnesota, St. Paul, Minnesota, USA

Robert Callender, Department of Biochemistry,Albert Einstein College of Medicine,Bronx, New York, USA

Britton Chance, Department of Biochemistry/Biophysics, University of Pennsylvania,Philadelphia, Pennsylvania, USA

Steven Chu, Lawrence Berkeley NationalLaboratory, Berkeley, California, USA

Louis J. DeFelice, Department of Pharmacology,Vanderbilt University, Nashville, Tennessee, USA

Johann Deisenhofer, Howard Hughes MedicalInstitute, The University of Texas, Dallas,Texas, USA

George Feher, Department of Physics,University of California, San Diego, La Jolla,California, USA

Hans Frauenfelder,Los Alamos National Laboratory,Los Alamos, New Mexico, USA

Ivar Giaever, Rensselaer Polytechnic Institute,Troy, New York, USA

Sol M. Gruner, Cornell University,Ithaca, New York, USA

Judith Herzfeld, Department of Chemistry,Brandeis University, Waltham, Massachusetts, USA

Mark S. Humayun, Doheny Eye Institute,Los Angeles, California, USA

Pierre Joliot, Institute de BiologiePhysico-Chimique, Fondation Edmondde Rothschild, Paris, France

Lajos Keszthelyi, Institute of Biophysics, HungarianAcademy of Sciences, Szeged, Hungary

Robert S. Knox, Department of Physicsand Astronomy, University of Rochester, Rochester,New York, USA

Aaron Lewis, Department of Applied Physics,Hebrew University, Jerusalem, Israel

Stuart M. Lindsay, Department of Physicsand Astronomy, Arizona State University,Tempe, Arizona, USA

David Mauzerall, Rockefeller University,New York, New York, USA

Eugenie V. Mielczarek, Department of Physicsand Astronomy, George Mason University, Fairfax,Virginia, USA

Markolf Niemz, Medical Faculty Mannheim,University of Heidelberg, Mannheim, Germany

V. Adrian Parsegian, Physical Science Laboratory,National Institutes of Health, Bethesda,Maryland, USA

Linda S. Powers, University of Arizona,Tucson, Arizona, USA

Earl W. Prohofsky, Department of Physics,Purdue University, West Lafayette, Indiana, USA

Andrew Rubin, Department of Biophysics, MoscowState University, Moscow, Russia

Michael Seibert, National Renewable EnergyLaboratory, Golden, Colorado, USA

David Thomas, Department of Biochemistry,University of Minnesota Medical School,Minneapolis, Minnesota, USA

Lorenz PavesiPhilippe M. Fauchet(Eds.)

Biophotonics

123

With 185 Figures

o

Professor Lorenzo PavesiUniversity of Trento, Department of Physics, Laboratory of NanoscienceVia Sommarive 14, 38050 Povo, ItalyE-mail: [email protected]

Philippe M. FauchetUniversity of Rochester, Department of Electrical and Computer Engineering160 Trustee Road, Rochester, NY 14627-0231, USAE-mail: [email protected]

c© 2008 Springer-Verlag Berlin Heidelberg

This work is subject to copyright. All rights are reserved, whether the whole or part of the materialis concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broad-casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication ofthis publication or parts thereof is permitted only under the provisions of the German Copyright Lawof September 9, 1965, in its current version, and permission for use must always be obtained fromSpringer-Verlag. Violations are liable to prosecution under the German Copyright Law.

The use of general descriptive names, registered names, trademarks, etc. in this publication does notimply, even in the absence of a specific statement, that such names are exempt from the relevantprotective laws and regulations and therefore free for general use.

Printed on acid-free paper

9 8 7 6 5 4 3 2 1

springer.com

ISBN 978-3-540-76779-4

Library of Congress Control Number: 2008920259

Cover: eStudio Calamar Steinen

e-ISBN 978-3-540-76782-4

Biological and Medical Physics, Biomedical Engineering ISSN 1618-7210

To AnnaTo Melanie

Preface

This book is the brainchild of the fourth Optoelectronic and Photonic WinterSchool on “Biophotonics” which took place from 25th February 2007 to 3rdMarch 2007 in Sardagna, a small village on the mountains around Trentoin Italy. This school, held every two years, have been promoted to trace thevery fast developing technologies and the tremendous progress in photonicswhich have occurred and will occur in the near future. It is a common viewthat its current explosive development will lead to deep paradigm shifts inthe near future. Identifying the plausible scenario for the future evolution ofphotonics presents an opportunity for constructive actions and scouting killertechnologies.

This book gathers together distinguished authors to give an overview of thelatest developments in biophotonics. Optical technologies have demonstrateda long tradition in the fields of life science and health care. The microscopehas opened for us a view into the world of cells and bacteria and has evolvedinto a modern powerful tool for basic biological research on cell processes.Modern surgical microscopes have become key tools in neurosurgery, as wellas in ophthalmology and surgery. Image-guided systems make use of computertomography and MRI data in navigated surgery. Endoscopes have openedan easy access to the inner parts of the human body in minimal invasivesurgery. Optical methods for gene sequencing and biochips open new routesfor diagnosis and treatment of cancer. Fluorescence methods have replacedradioactive methods in screening processes in drug development. The role ofoptical technologies as the key enabling factor in health care and life scienceswill grow tremendously in future. It is therefore a good time to write a bookon biophotonics with the aim to introduce students and young researchers tothis field and make a point of the importance of this science and technology.In particular, each chapter consists of a review, an introduction to the subjectand a presentation of the current state of the art.

The first three chapters of the book are by Bassi, Giacometti and cowork-ers on photosynthesis, on the application of photosynthesis to biofuel produc-tion and on the role of carotenoids in photoprotection. Then, the chapter by

VIII Preface

Diaspro and coworkers introduces the readers to non-linear microscopy. An-other kind of microscopy, spectral self-interference microscopy, is presented inthe chapter by Unlu and coworkers. The following chapter, also by Unlu andcoworkers, deals with resonant cavity biosensors for sensing and imaging. Adifferent way to engineer biosensors is the use of optical microcavities: theseare reviewed in the chapter by Fauchet and coworkers.

The success and application of optical coherence tomography are describedin the chapter by Boppart. Other coherent laser techniques for medical diag-nostics are described in the chapter by Kemper and von Bally.

Two chapters follow on the use of fluorescence labelling for sensing andquantitative analysis. Caiolfa and coworkers detail both the principle andapplications of luminescence probes in quantitative biology. Ligler reviewsfluorescent-based optical biosensors.

The next chapter, by Seitz, discusses optical biochips and their applicationsto lab-on-a-chip. An example of the merging of microelectronics with biologyis given in the chapter by Charbon on the use of CMOS camera for bioimagingapplications.

Going back to basic optics, the chapter by Chiou and coworkers shows theuse of optical forces in trapping and manipulating bio-objects, such as redblood cells or proteins. Optics can also be used in surgery. Pini and coworkersdiscuss the application of lasers in ophthalmology and vascular surgery intheir chapter.

The final two chapters are examples of how photonics is used in commonclinical practice. In the chapter by Pentland, the photobiology of the skin andthe phototherapy are discussed, while in the chapter by Wilson the use oflight in therapy is introduced and a few examples are given in detail.

We thank all the authors who presented very interesting state-of-the artlectures and chapters. Last but not least, we express our gratitude also toall those who have contributed to the success of the fourth Optoelectronicand Photonics Winter School on Biophotonics: the organizing committee (L.Pavesi, P. Fauchet, M. Anderle, L. Gonzo, R. Antolini, M. Ferrari, S. Iannotta,G. Righini), the staff of the University of Trento and all the students, whoseparticipation was lively and stimulating. This year the winter school was alsothe first edition of the AIV school on nanotechnology. We acknowledge thegenerous contribution by the Societa Italiana di Scienze e Tecnologie throughits president Mariano Anderle. Further support came from the University ofTrento, the Department of Physics, FBK-irst, LaserOptronic and Spectra-Physics, Crisel-Instruments and LOT Oriel group Europe. Particular thanksare also due to Alessandro Pitanti and his brother for their help in the editingthe Latex files.

Trento, Rochester Lorenzo PavesiMay 2008 Philippe M. Fauchet

Contents

1 Light Conversion in Photosynthetic OrganismsS. Frigerio, R. Bassi, and G.M. Giacometti . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Chloroplast Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Pigments and Light Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4 Photosynthetic Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.4.1 Photosystem II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.4.2 Photosystem I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.4.3 Cytochrome b6f . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.4.4 ATP Synthase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.5 Cyclic Phosphorylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.6 Photoinhibition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Exploiting Photosynthesis for Biofuel ProductionC. Govoni, T. Morosinotto, G. Giuliano, and R. Bassi . . . . . . . . . . . . . . . 152.1 Biological Production of Vehicle Traction Fuels: Bioethanol

and Biodiesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.1.1 Bioethanol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.1.2 Biodiesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.1.3 Biofuels Still Present Limitations Preventing

Their Massive Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.2 Hydrogen Biological Production by Fermentative Processes . . . . . . . 19

2.2.1 Hydrogen Production by Bacterial Fermentation . . . . . . . . . 202.3 Hydrogen Production by Photosynthetic Organisms . . . . . . . . . . . . . 21

2.3.1 Cyanobacteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.3.2 Eukaryotic Algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.4 Challenges in Algal Hydrogen Production . . . . . . . . . . . . . . . . . . . . . . 232.4.1 Oxygen Sensitivity of Hydrogen Production . . . . . . . . . . . . . . 232.4.2 Optimization of Light Harvesting in Bioreactors . . . . . . . . . . 25

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

X Contents

3 In Between Photosynthesis and Photoinhibition:The Fundamental Role of Carotenoids and Carotenoid-BindingProteins in PhotoprotectionG. Bonente, L. Dall’Osto, and R. Bassi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.1 When Light Becomes Dangerous for a Photosynthetic Organism . . 293.2 Acclimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.3 State 1–State 2 Transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323.4 Carotenoids Play a Fundamental Role in Many Photoprotection

Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.5 Analysis of Xanthophyll Function In Vivo . . . . . . . . . . . . . . . . . . . . . . 363.6 Nonphotochemical Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.7 Feedback Deexcitation of Singlet-Excited Chlorophylls: qE . . . . . . . 393.8 ∆pH - Independent Energy Thermal Dissipation (qI) . . . . . . . . . . . . 403.9 Chlorophyll Triplet Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.10 Scavenging of Reactive Oxygen Species . . . . . . . . . . . . . . . . . . . . . . . . 413.11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4 Non-Linear MicroscopyD. Mazza, P. Bianchini, V. Caorsi, F. Cella, P.P. Mondal, E. Ronzitti,I. Testa, G. Vicidomini, and A. Diaspro . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474.2 Chronological Notes on MPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.3 Principles of Confocal and Two-Photon Fluorescence Microscopy . . 49

4.3.1 Fluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.3.2 Confocal Principles and Laser Scanning Microscopy . . . . . . . 504.3.3 Point Spread Function of a Confocal Microscope . . . . . . . . . 52

4.4 Two-Photon Excitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.5 Two-Photon Optical Sectioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594.6 Two-Photon Optical Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604.7 Second Harmonic Generation (SHG) Imaging . . . . . . . . . . . . . . . . . . . 634.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

5 Applications of Optical Resonance to Biological Sensingand Imaging: I. Spectral Self-Interference MicroscopyM.S. Unlu, A. Yalc. in, M. Dogan, L. Moiseev, A. Swan, B.B. Goldberg,and C.R. Cantor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 715.1 High-Resolution Fluorescence Imaging . . . . . . . . . . . . . . . . . . . . . . . . . 715.2 Self-Interference Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 715.3 Physical Model of SSFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.3.1 Classical Dipole Emission Model . . . . . . . . . . . . . . . . . . . . . . . 735.4 Acquisition and Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

5.4.1 Microscope Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 755.4.2 Fitting Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Contents XI

5.5 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 775.5.1 Monolayers of Fluorophores on Silicon Oxide Surfaces:

Fluorescein, Quantum Dots, Lipid Films . . . . . . . . . . . . . . . . 775.5.2 Conformation of Surface-Immobilized DNA . . . . . . . . . . . . . . 79

5.6 SSFM in 4Pi Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 825.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

6 Applications of Optical Resonance to Biological Sensingand Imaging: II. Resonant Cavity BiosensorsM.S. Unlu, E. Ozkumur, D.A. Bergstein, A. Yalc. in, M.F. Ruane,and B.B. Goldberg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 876.1 Multianalyte Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 876.2 Resonant Cavity Imaging Biosensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

6.2.1 Detection Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 886.2.2 Experimental Setup, Data Acquisition, and Processing . . . . 906.2.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 916.2.4 Spectral Reflectivity Imaging Biosensor . . . . . . . . . . . . . . . . . 92

6.3 Optical Sensing of Biomolecules Using Microring Resonators . . . . . 946.3.1 Basics on Microring Resonators . . . . . . . . . . . . . . . . . . . . . . . . 946.3.2 Setup and Data Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 956.3.3 Data Analysis and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 96

6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

7 Biodetection Using Silicon Photonic Crystal MicrocavitiesP.M. Fauchet, B.L. Miller, L.A. DeLouise, M.R. Lee, and H. Ouyang . . 1017.1 Photonic Crystals: A Short Introduction . . . . . . . . . . . . . . . . . . . . . . . 101

7.1.1 Electromagnetic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1017.1.2 One-Dimensional and Two-Dimensional PhC . . . . . . . . . . . . . 1037.1.3 Microcavities: Breaking the Periodicity . . . . . . . . . . . . . . . . . . 1057.1.4 Computational Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

7.2 One-Dimensional PhC Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1077.2.1 Preparation and Selected Properties of Porous Silicon . . . . . 1077.2.2 Sensing Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1097.2.3 One-Dimensional Biosensor Design and Performance . . . . . 1117.2.4 Fabrication of One-Dimensional PhC Biosensors . . . . . . . . . . 112

7.3 Selected Biosensing Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1147.3.1 DNA Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1147.3.2 Bacteria Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1147.3.3 Protein Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1157.3.4 IgG Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

7.4 Two-Dimensional PhC Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1187.4.1 Sample Preparation and Measurement . . . . . . . . . . . . . . . . . . 1187.4.2 Sensing Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1197.4.3 Selected Biosensing Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

XII Contents

7.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

8 Optical Coherence Tomography with Applicationsin Cancer ImagingS.A. Boppart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1278.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1278.2 Principles of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1288.3 Optical Sources for Optical Coherence Tomography . . . . . . . . . . . . . 1338.4 Fourier-Domain Optical Coherence Tomography . . . . . . . . . . . . . . . . 1338.5 Beam Delivery Instruments for Optical Coherence Tomography . . . 1358.6 Spectroscopic Optical Coherence Tomography . . . . . . . . . . . . . . . . . . 1368.7 Applications to Cancer Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

8.7.1 Cellular Imaging for Tumor Cell Biology . . . . . . . . . . . . . . . . 1388.7.2 Translational Breast Cancer Imaging . . . . . . . . . . . . . . . . . . . . 140

8.8 Optical Coherence Tomography Contrast Agents . . . . . . . . . . . . . . . . 1418.9 Molecular Imaging using Optical Coherence Tomography . . . . . . . . . 1458.10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

9 Coherent Laser Measurement Techniquesfor Medical DiagnosticsB. Kemper and G. von Bally . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1519.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1519.2 Electronic Speckle Pattern Interferometry (ESPI) . . . . . . . . . . . . . . . 152

9.2.1 Double Exposure Subtraction ESPI . . . . . . . . . . . . . . . . . . . . . 1529.2.2 Spatial Phase Shifting (SPS) ESPI . . . . . . . . . . . . . . . . . . . . . 153

9.3 Endoscopic Electronic Speckle Pattern Interferometry (ESPI) . . . . . 1569.3.1 Proximal Endoscopic ESPI . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1569.3.2 Distal Endoscopic ESPI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158

9.4 Microscopic (Speckle) Interferometry . . . . . . . . . . . . . . . . . . . . . . . . . . 1619.5 Digital Holographic Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

9.5.1 Principle and Measurement Setup . . . . . . . . . . . . . . . . . . . . . . 1649.5.2 Nondiffractive Reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . 1669.5.3 Resolution and Numerical Focus . . . . . . . . . . . . . . . . . . . . . . . . 1709.5.4 Digital Holographic Phase Contrast Microscopy

of Living Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1719.6 Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

10 Biomarkers and Luminescent Probes in QuantitativeBiologyM. Zamai, G. Malengo, and V.R. Caiolfa . . . . . . . . . . . . . . . . . . . . . . . . . . . 17710.1 Fluorophores and Genetic Dyes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177

10.1.1 Small Organic Dyes and Quantum Dots . . . . . . . . . . . . . . . . . 17710.1.2 Fluorescent Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

Contents XIII

10.2 Microspectroscopy in Quantitative Biology: Where and How . . . . . . 18310.2.1 Fluorescence Correlation Spectroscopy . . . . . . . . . . . . . . . . . . 18310.2.2 Fluorescence Lifetime Imaging (FLIM) . . . . . . . . . . . . . . . . . . 18810.2.3 Glossary of Molecular Biology . . . . . . . . . . . . . . . . . . . . . . . . . . 194

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

11 Fluorescence-Based Optical BiosensorsF.S. Ligler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19911.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19911.2 Biological Recognition Molecules and Assay Formats . . . . . . . . . . . . 20011.3 Displacement Immunosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20311.4 Fiber Optic Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

11.4.1 Fiber Optics for Biosensor Applications . . . . . . . . . . . . . . . . . 20511.4.2 Optrode Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20711.4.3 Evanescent Fiber Optic Biosensors . . . . . . . . . . . . . . . . . . . . . . 207

11.5 Bead-Based Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20911.6 Planar Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21011.7 Critical Issues and Future Opportunities . . . . . . . . . . . . . . . . . . . . . . . 212References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214

12 Optical BiochipsP. Seitz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21712.1 Taxonomy of Optical Biochips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

12.1.1 Basic Architecture of Optical Biochips . . . . . . . . . . . . . . . . . . 21712.2 Analyte Classes for Optical Biochips . . . . . . . . . . . . . . . . . . . . . . . . . . 220

12.2.1 DNA (DNA Fragments, mRNA, cDNA) . . . . . . . . . . . . . . . . . 22012.2.2 Proteins (Antigens) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22112.2.3 Specific Organic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22112.2.4 Cell Gene Products (cDNA, Proteins) . . . . . . . . . . . . . . . . . . . 22112.2.5 Tissue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

12.3 Optical Effects for Biochemical Sensors . . . . . . . . . . . . . . . . . . . . . . . . 22212.3.1 Spectral Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22212.3.2 Phase Shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22312.3.3 Fluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22312.3.4 Luminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22312.3.5 Raman Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22412.3.6 Nonlinear Optical (NLO) Effects . . . . . . . . . . . . . . . . . . . . . . . 224

12.4 Preferred Sensing Principles for Optical Biochips . . . . . . . . . . . . . . . . 22412.4.1 Evanescent Wave Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22512.4.2 Fluorescence Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

12.5 Readout Methods for Evanescent Wave Sensors . . . . . . . . . . . . . . . . . 22912.5.1 Angular Scanning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22912.5.2 Wavelength Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23012.5.3 Grating Coupler Chirping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

12.6 Substrates for Optical Biochips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23012.7 Realization Example of an Optical Biosensor/Biochip: WIOS . . . . . 231

XIV Contents

12.8 Outlook: Lab-on-a-Chip Using Organic Semiconductors . . . . . . . . . . 23212.8.1 Basics of Organic Semiconductors . . . . . . . . . . . . . . . . . . . . . . 23312.8.2 Organic LEDs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23312.8.3 Organic Lasers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23412.8.4 Organic Photodetectors and Image Sensors . . . . . . . . . . . . . . 23412.8.5 Organic Photovoltaic Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23412.8.6 Organic Field Effect Transistors and Circuits . . . . . . . . . . . . 23512.8.7 Monolithic Photonic Microsystems Using Organic

Semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23512.9 Conclusions and Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236

13 CMOS Single-Photon Systems for Bioimaging ApplicationsE. Charbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23913.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23913.2 Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24013.3 Lifetime Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24213.4 Time-of-Flight in Bio- and Medical Imaging . . . . . . . . . . . . . . . . . . . . 24413.5 System Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24513.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248

14 Optical Trapping and Manipulation for BiomedicalApplicationsA. Chiou, M.-T. Wei, Y.-Q. Chen, T.-Y. Tseng, S.-L. Liu,A. Karmenyan, and C.-H. Lin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24914.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24914.2 Theoretical Models for the Calculation of Optical Forces . . . . . . . . . 252

14.2.1 The Ray-Optics (RO) Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 25214.2.2 Electromagnetic (EM) Model . . . . . . . . . . . . . . . . . . . . . . . . . . 255

14.3 Experimental Measurements of Optical Forces . . . . . . . . . . . . . . . . . . 25514.3.1 Axial Optical Force as a Function of Position

along the Optical Axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25514.3.2 Transverse Trapping Force Measured by Viscous Drag . . . . . 25714.3.3 Three-Dimensional Optical Force Field Probed

by Particle Brownian Motion . . . . . . . . . . . . . . . . . . . . . . . . . . 25714.3.4 Optical Forced Oscillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261

14.4 Potential Biomedical Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26514.4.1 Optical Forced Oscillation for the Measurement

of Protein–Protein Interactions . . . . . . . . . . . . . . . . . . . . . . . . . 26614.4.2 Protein–DNA Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26714.4.3 Optical Trapping and Stretching of Red Blood Cells . . . . . . 269

14.5 Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

Contents XV

15 Laser Tissue Welding in Minimally Invasive Surgeryand MicrosurgeryR. Pini, F. Rossi, P. Matteini, and F. Ratto . . . . . . . . . . . . . . . . . . . . . . . . 27515.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27515.2 Laser Welding in Ophthalmology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281

15.2.1 Laser Welding of the Cornea . . . . . . . . . . . . . . . . . . . . . . . . . . . 28115.2.2 Combing Femtosecond Laser Microsculpturing

of the Cornea with Laser Welding . . . . . . . . . . . . . . . . . . . . . . 28515.2.3 Laser Closure of Capsular Tissue . . . . . . . . . . . . . . . . . . . . . . . 287

15.3 Applications in Microvascular Surgery . . . . . . . . . . . . . . . . . . . . . . . . . 28915.4 Potentials in Other Surgical Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291

15.4.1 Laser Welding of the Gastrointestinal Tract . . . . . . . . . . . . . . 29115.4.2 Laser Welding in Gynaecology . . . . . . . . . . . . . . . . . . . . . . . . . 29115.4.3 Laser Welding in Neurosurgery . . . . . . . . . . . . . . . . . . . . . . . . . 29215.4.4 Laser Welding in Orthopaedic Surgery . . . . . . . . . . . . . . . . . . 29215.4.5 Laser Welding of the Skin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29215.4.6 Laser Welding in Urology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293

15.5 Perspectives of Nanostructured Chromophores for Laser Welding . . 293References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296

16 Photobiology of the SkinA.P. Pentland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30116.1 Basics of Skin Structure: Cell Types, Skin Structures,

and Their Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30116.2 Effects of Light Exposure on Skin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30416.3 Sun Protection and Sunscreens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30916.4 Phototherapy: Use of Light for Treatment for Skin Disease . . . . . . . 311References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313

17 Advanced Photodynamic TherapyB.C. Wilson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31517.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31517.2 Basic Principles and Features of “Standard PDT” . . . . . . . . . . . . . . . 31617.3 Novel PDT Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319

17.3.1 Two-Photon PDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31917.3.2 Metronomic PDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32217.3.3 PDT Molecular Beacons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32317.3.4 Nanoparticle-Based PDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325

17.4 PDT Dosimetry Using Photonic Techniques . . . . . . . . . . . . . . . . . . . . 32717.5 Biophotonic Techniques for Monitoring Response to PDT . . . . . . . . 33017.6 Biophotonic Challenges and Opportunities in Clinical PDT. . . . . . . 33117.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335

List of Contributors

Gert von BallyCenter for Biomedical Opticsand Photonics,University of Munster,Robert-Koch-Straße 45,48129 Munster,[email protected]

Roberto BassiDepartement de Biologie - Case 901,Faculte des Sciences de Luminy,Universite Aix-Marseille II, LGBP,163, Avenue de Luminy13288 Marseille Cedex 09,FranceandDipartimento Scientificoe Tecnologico,Universita di VeronaStrada Le Grazie 15,37134 Verona,[email protected]

David A. BergsteinDepartment of Electricaland Computer Engineering,Boston University,8 St. Mary’s St.,02215 Boston, MA, [email protected]

Paolo BianchiniLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, Italy

Giulia BonenteLaboratoire de Genetique et deBiophysique des Plantes,Faculte de Sciences de Luminy,Universite de Marseille,Marseille Cedex 9,[email protected]

andDipartimento Scientifico eTecnologico, Universita di Verona,Strada le Grazie 15,37134 Verona,Italy

Stephen A. BoppartBeckman Institute for AdvancedScience and Technology,Departments of Electricaland Computer Engineering,Bioengineering, and MedicineUniversity of Illinois atUrbana-ChampaignMills Breast Cancer Institute

XVIII List of Contributors

Carle Foundation HospitalUrbana, IL [email protected]

Valeria R. CaiolfaDepartment of Molecular Biologyand Functional Genomics,San Raffaele Scientific InstituteMilano, [email protected]

andIIT Network Research,Unit of Molecular Neuroscience,San Raffaele Scientific Institute,Milano, Italy

Charles R. CantorDepartment of BiomedicalEngineering, 44 Cummington St.,02215 Boston, MA, [email protected]

andCenter for Advanced Biotechnology,36 Cummington St.,02215 Boston, MA, USA

Valentina CaorsiLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, Italy

Francesca CellaLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, Italy

Edoardo CharbonQuantum Architecture Group(AQUA), EPFL,1015 Lausanne, [email protected]

Yin-Quan ChenInstitute of BiophotonicsEngineering, National Yang-MingUniversity Taipei, Taiwan, ROC

Arthur ChiouInstitute of BiophotonicsEngineering, National Yang-MingUniversity Taipei, Taiwan, [email protected]

Luca Dall’OstoDipartimento Scientifico eTecnologico, Universita di Verona,Strada le Grazie 15,37134 Verona,[email protected]

Lisa A. DeLouis eCenter for Future Health,University of Rochester,Rochester, NY, USAandDepartment of Dermatology,University of RochesterRochester, NY,USA

Alberto DiasproLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa,Italy

Mehmet DoganDepartment of Physics,Boston University,590 Commonwealth Ave.,02215 Boston, MA, [email protected]

Philippe M. FauchetCenter for Future Health,University of Rochester,Rochester, NY, USAand

List of Contributors XIX

Department of Electricaland Computer Engineering,University of Rochester,Rochester, NY, USAandDepartment of BiomedicalEngineering,University of Rochester,Rochester, NY, USAandThe Institute of Optics,University of Rochester,Rochester, NY, USA

Sara FrigerioUniversite Aix-Marseille II, LGBP,Faculte des Sciences de Luminy,Departement de Biologie - Case 901,163, Avenue de Luminy13288 Marseille Cedex 09,France

Giorgio GiacomettiDipartimento di Biologia,Universita di Padova,Via Ugo Bassi 58 B,35131 Padova, Italy

Giovanni M. GiulianoEnte per le Nuove tecnologie,l’Energia e l’Ambiente (ENEA),Unita Biotecnologie,Centro Ricerche Casaccia,C.P. 2400, Roma 00100,Italy

Bennett B. GoldbergDepartment of Physics,Boston University,590 Commonwealth Ave.,02215 Boston, MA,[email protected]

and

Department of Electricaland Computer Engineering,Boston University,8 St. Mary’s St.,02215 Boston, MA, USAandDepartment of BiomedicalEngineering, 44 Cummington St.,02215 Boston, MA, USA

Chiara GovoniDipartimento Scientifico eTecnologico, Universita di Verona,Strada Le Grazie 15,37134 Verona, Italy

Artashes KarmenyanInstitute of BiophotonicsEngineering, National Yang-MingUniversity Taipei, Taiwan, ROC

Bjorn KemperCenter for Biomedical Opticsand Photonics,University of Munster,Robert-Koch-Straße 45,48129 Munster, [email protected]

Mindy R. LeeCenter for Future Health,University of Rochester,Rochester, NY, USAandThe Institute of Optics,University of Rochester,Rochester, NY, USA

Frances S. LiglerCenter for Bio/Molecular Science& Engineering,Naval Research Laboratory,Washington, DC 20375, [email protected]

XX List of Contributors

Chi-Hung LinInstitute of BiophotonicsEngineering, National Yang-MingUniversity Taipei, Taiwan, ROC

Shang-Ling LiuInstitute of BiophotonicsEngineering, National Yang-MingUniversity Taipei, Taiwan, ROC

Gabriele MalengoDepartment of Molecular Biologyand Functional Genomics,San Raffaele Scientific Institute,Milano, Italy

Paolo MatteiniIstituto di Fisica Applicata,Consiglio Nazionale delle RicercheVia Madonna del Piano 10,50019 Sesto Fiorentino,Italy

Davide MazzaLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, [email protected]

Benjamin L. MillerCenter for Future Health,University of Rochester,Rochester, NY, USAandDepartment of BiomedicalEngineering, University of Rochester,Rochester, NY, USAandDepartment of Dermatology,University of RochesterRochester, NY, USA

Lev MoiseevCenter for Advanced Biotechnology36 Cummington St., 02215 Boston,MA, [email protected]

Partha P. MondalLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, Italy

Tomas MorosinottoDipartimento di Biologia,Universita di Padova,Via Ugo Bassi 58 B,35131 Padova, Italy

Huimin OuyangDepartment of Electricaland Computer Engineering,University of Rochester,Rochester, NY, USA

Emre OzkumurDepartment of Electrical andComputer Engineering,Boston University,8 St. Mary’s St.,02215 Boston, MA, [email protected]

Alice P. PentlandDepartment of Dermatology,University of Rochester,601 Elmwood,Rochester, NY 14642, USAAlice [email protected]

Roberto PiniIstituto di Fisica Applicata,Consiglio Nazionale delle RicercheVia Madonna del Piano 10,50019 Sesto Fiorentino, [email protected]

List of Contributors XXI

Fulvio RattoIstituto di Fisica Applicata,Consiglio Nazionale delle Ricerche,Via Madonna del Piano 10,50019 Sesto Fiorentino, Italy

Emiliano RonzittiLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, Italy

Francesca RossiIstituto di Fisica Applicata,Consiglio Nazionale delle RicercheVia Madonna del Piano 10,50019 Sesto Fiorentino, Italy

Michael F. RuaneDepartment of Electricaland Computer Engineering,Boston University,8 St. Mary’s St.,02215, Boston, MA, [email protected]

Peter SeitzSwiss Center for Electronicsand Microtechnology,CSEM SA,Zurich Switzerland,andInstitute for Microtechnology,University of Neuchatel,Neuchatel, [email protected]

Anna SwanDepartment of Electricaland Computer Engineering,Boston University,8 St. Mary’s St.,02215 Boston, MA, [email protected]

Ilaria TestaLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa, Italy

Te-Yu TsengInstitute of BiophotonicsEngineering, National Yang-MingUniversity, Taipei, Taiwan, ROC

M. Selim UnluDepartment of Electricaland Computer Engineering,Boston University,8 St. Mary’s St.,02215 Boston, MA,[email protected]

andDepartment of Physics,Boston University,590 Commonwealth Ave.,02215 Boston, MA, USAandDepartment of BiomedicalEngineering,44 Cummington St.,02215 Boston, MA,USA

Giuseppe VicidominiLAMBS-MicroSCoBiO ResearchCenter, Department of Physics,University of Genoa,Via Dodecaneso 33,16146 Genoa,Italy

Ming-Tzo WeiInstitute of BiophotonicsEngineering, National Yang-MingUniversity, Taipei, Taiwan, ROC

XXII List of Contributors

Brian C. WilsonDivision of Biophysics andBioimaging, Ontario CancerInstitute, Toronto,ON, CanadaandDepartment of Medical Biophysics,University of Toronto, Toronto,ON, Canada, M5G [email protected]

Ayca Yalc.inDepartment of Electricaland Computer Engineering,Boston University,

8 St. Mary’s St.,02215 Boston, MA, [email protected]

Moreno ZamaiDepartment of Molecular Biologyand Functional Genomics,San Raffaele Scientific Institute,Milano, ItalyandIIT Network Research,Unit of Molecular Neuroscience,San Raffaele Scientific Institute,Milano, [email protected]

1

Light Conversion in Photosynthetic Organisms

S. Frigerio, R. Bassi, and G.M. Giacometti

1.1 Introduction

The sun is the ultimate source of free energy that drives all of the processesin living cells. The radiant energy of the sun is captured and converted tochemical energy by photosynthesis. The flux of carbon through the biospherebegins with photosynthesis. Photosynthetic organisms produce carbohydratesand molecular oxygen from carbon dioxide and water:

6H2O + 6CO2 + light → (CH2O)6 pedice + 6O2 (1.1)

The carbohydrates produced by photosynthesis serve as the energy source forother non-photosynthetic (heterotrophic) organisms. In this process, carbo-hydrates are recycled to carbon dioxide and water by the combined action ofcellular catabolic processes.

The fixation of carbon dioxide into sugars requires free energy in the formof ATP and reducing power in the form of NADPH. The light reactions ofphotosynthesis respond to this need: the visible component of solar radiation iscaptured and its energy is converted into ATP and NADPH through a complexseries of redox reactions and membrane-mediated energy conversions.

The following reactions, referred to as dark phase of photosynthesis, as,in principle, they do not need solar radiation to be carried out if NADPHand ATP are provided, drive the reduction of CO2 to the carbohydrate GAP(glyceraldehyde-3-phosphate). The series of reactions are altogether indicatedas the Calvin-Benson cycle [1].

The enzyme directly responsible for the addition of CO2 to ribulose 1,5 -bisphosphate is RubisCO [2]. The whole carboxylation cycle requires threeCO2 molecules to generate a molecule of GAP, which is then used for thesynthesis of more complex sugars or other compounds. The energy requiredfor this cycle is nine molecules of ATP and six of NADPH, as summarized inthe following equation:

3CO2 + 9ATP + 6NADPH → GAP + 9ADP + 8Pi + 6NADP+ (1.2)

2 S. Frigerio et al.

To carry out these reactions, in particular those of the light phase, plantsneed a specific apparatus, which is in part shared by other photosyntheticorganisms like cyanobacteria and green algae. The photosynthetic apparatusis located within the cells in dedicated organelles called chloroplasts. Theorigins of chloroplasts is most probably due to an event of endosymbiosis of acyanobacterium by an eukaryotic cell [3].

1.2 Chloroplast Structure

Chloroplasts are characterized by two different membrane systems: a doublemembrane envelope, which encloses a soluble fraction called stroma; withinthe stroma are located thylakoids, an extended and morphologically complexmembrane system carrying the photosynthetic apparatus. This inner membra-nous system defines an internal space called thylakoid lumen. The membranecan be divided in two main regions: the grana region, where the membranesform stacked structures of flat vesicles and the stroma lamellae, which createconnections between stacked vesicles, insuring continuity to the lumenal space(Fig. 1.1). Thylakoid membranes are composed principally of glycerol-lipids,in particular monogalactosyldiacyl-glycerol and digalactosyldiacyl-glycerol,which distinguish photosynthetic membranes from all the others [4].

The prokaryotic origin of chloroplast is supported by the presence of acircular genome, which encodes almost 90–120 sequences, while many others(∼3,400) or more proteins, predicted as chloroplastic, are encoded by nucleargenes [5].

Fig. 1.1. Chloroplast structure. Localization in a typical mesophyll cell, schematicrepresentation and electron micrograph, showing thylakoid membranes divided ingrana and stroma lamellae

1 Light Conversion in Photosynthetic Organisms 3

1.3 Pigments and Light Absorption

The capacity of plants to absorb light covers a wide range of wavelengths,between 350 and 700 nm, which include almost 60% of incident sunlightspectrum. All the pigments involved in light absorption, which are groupedin to two big classes, chlorophylls and carotenoids, are located within thethylakoids. Chlorophylls, named a or b, because of two different chemicalforms, are organo-metallic compounds in which a substituted porphyrin ringstructure coordinates a magnesium atom in its center. A phytol chain, neces-sary for their localization within the lipid membrane, constitutes one of thelateral chains of the porphyrin. These pigments strongly absorb red and bluelight, while they scatter green light, thus giving reason for plant leafs color [6].

The absorption of a photon drives chlorophylls to an excited state, con-verting electromagnetic energy into electronic excitation; red light determinesexcitation to the first singlet excited state, while blue light causes excitationto the second singlet state (Fig. 1.2).

The photosynthetic apparatus makes the job of converting such molecularexcitation into different forms of stable free energy. Electronic excitation ofmolecules decays to ground state in the nanosecond time scale, and the reac-tions that convert the excitation energy must be fast enough to successfullycompete with decay.

The higher singlet-excited states, as well as the excited vibrational states,decay by internal conversion to the lowest vibrational state of the lowest ex-cited singlet, in the time range of femtosecond, which is much faster than the

Abs

orpt

ion

of b

lue

light

Abs

orpt

ion

of r

ed li

ght

Flu

ores

cenc

e

Rel

axat

ion

Energy

Rel

axat

ion

Rel

axat

ion

1st excited singlet state

2nd excited singlet state

Ground state

Fluorescence

Absorption

Red

Blue

Fig. 1.2. Light absorption and energy decay scheme


rate of any subsequent process; therefore, the system equilibrate at the firstsinglet-excited state before anything else can happen.

Relaxation to ground state may occur through different pathways: (1)internal conversion, where the excitation energy is dissipated as heat; (2)fluorescence emission, where the energy is emitted as a photon; (3) energytransfer, where excitation energy is transferred to a nearby acceptor molecule(Forster energy transfer) [7]. In the photosynthetic apparatus, the most im-portant is the latter: reiterated energy transfer between neighbor chlorophyllmolecules allows excitation energy to reach a particular site (reaction center),where a redox reaction takes place, converting the excitation into chemicalenergy.

For efficient conversion of light energy into chemical energy, a highly effi-cient energy transfer through neighboring pigments is essential. However, theideal conditions for high efficiency in energy transfer is not always satisfiedin the photosynthetic apparatus; if the energy transfer rate is not competingsuccessfully with the other relaxation mechanisms, the excitation energy islost and a fluorescence emission or heat dissipation takes place.

Another important phenomenon must be taken into account. The lowestsinglet-excited state of chlorophylls can also undergo intersystem crossing tothe triplet state. This excited state cannot transfer its energy to the reactioncenter and therefore cannot contribute to photochemistry. Worse than that,chlorophyll triplet state can easily activate molecular oxygen to its singletstate, producing a highly reactive and dangerous molecule.

The second class of pigments is represented by carotenoids, linear poly-enes [8] involved in facing the problem of singlet oxygen production, besidescontributing to absorb light and transfer it to the nearby chlorophylls: theycan quench chlorophyll triplet-excited states and also scavenge singlet oxy-gen. In both cases, this leads to the dissipation of the absorbed photon butit prevents oxidative damage to the membrane and photosynthetic apparatuscomponents.

1.4 Photosynthetic Apparatus

Light absorption and energy conversion take place in specific and highly or-ganized structures embedded in the thylakoid membrane. All together theyconstitute the photosynthetic apparatus. There are four different complexes:two Photosystems (PSI and PSII), the cytochrome b6f, and the F-ATPase.These are composed of several protein subunits, cofactors, and pigments, whileadditional electron carriers can move within the lipid bilayer of the mem-brane or in the aqueous medium (Fig. 1.3). In the last years, high resolutionstructures for the four complexes, even though from different species, havebecome available, thus allowing a deeper comprehension of photosyntheticmechanisms [9–13].


Fig. 1.3. Photosynthetic apparatus: Both the photosystems, cytochrome b6f, andF-ATPase are shown, in addition to movable subunits as plastoquinone (PQ), plas-tocyanin (PC), and ferredoxin (Fdx)

As pointed out earlier, the task of the photosynthetic apparatus is thatof providing reducing power and ATP to drive CO2 fixation. The reducingpower in the form of NADPH is generated through a redox reaction in whichwater is the ultimate electron donor:

2NADP+ + 2H2O + light → 2NADPH + O2 + 2H+ (1.3)

The energy of a single photon is not sufficient to drive an electron from waterto NADP+, and two photons must be used to make the job. For this reason,organisms using water as electron donor, i.e., oxygenic organisms such ashigher plants, algae and cyanobacteria, make use of two photosystems thatcan operate in series.

Photosystem II is able to extract electrons from water and brings themto plastoquinone, a quinone molecule freely diffusible in the membrane lipidmoiety. Photosystem I makes use of another photon to bring the electron atthe level of NADP+.

The connection between the two photosystems is operated by cytochromeb6f complex, which transfers electrons from the plastoquinone pool to plasto-cyanine, a soluble Cu-protein freely diffusible in the thylakoids lumen. Thiselectron transfer is in favor of potential gradient, and there is no need of lightenergy to drive it. Actually, the role of this complex is not limited to simplycatalyze an electron transfer, as this would lead to a significant loss of energy.On the contrary, cytochrome b6f, using the electron transfer energy, acts asa proton pump, which transfers protons against gradient from the stromal tothe lumenal side of the membrane. The action of cytochrome b6f is all theway analogue to that of complex III (or cytochrome b1c) of the mitochondrial


−1.5

−1.0

−0.5

0

Oxygen-evolvingcomplex

Cytochrome bf complex Plastocyanin

Light

8H

PSII

Red

uction

pot

ential

(V

)

P680*

P700

P700*

PSI

Light

2NADPH

2H

Fd-NADPoxidoreductase

2NADP

Fd

FB

FA

FX

A1

A0

P680

PQpoolPQB

PQA

2H2O

4H

O2

++

Z

Pha

+0.5

+1.0

+1.5

+

+

+ +

Fig. 1.4. Electron transport chain

respiratory chain, and plays an essential role in building up the proton gradientnecessary to drive ADP phosphorylation by ATP synthase (Fig. 1.4). Photo-systems are composed of proteins and pigments, chlorophylls and carotenoids,responsible for light harvesting; PSI and PSII are different in several aspects,in particular for the supramolecular organization, but it is possible to identifythe same two functional moieties in both: (a) a core complex, responsible forcharge separation and the first steps of the electron transport and (b) an an-tenna complex responsible of increasing the light harvesting and transferringof absorbed energy to the reaction center.

Both the core complexes are composed of protein encoded by psa and psbgenes, respectively, for PSI and PSII, which bind pigments and surround the socalled “special chlorophyll pair,” where the charge separation actually occurs.

The core complex is surrounded by antenna proteins, encoded by the multi-genic family of light-harvesting complexes (lhc), which primarily increase thelight-harvesting capacity, but are also involved in regulation and photopro-tection [14]. PSI antenna complex is composed of Lhca1-4 as major subunitsand Lhca5-6 as minor [15]; on the contrary, the major component of PSIIantenna complex is a heterotrimer of Lhcb1-3 subunits, while remaining Lhcbproteins, Lhcb4-6, are altogether classified as minor antennas and they aregenerally found as monomers [16].

1.4.1 Photosystem II

Photosystem II (PSII), which is mainly located in grana thylakoids, is alight driven water-plastoquinone reductase. It performs the thermodynami-cally most demanding and most dangerous reaction: splitting water into itselementary components, molecular oxygen and protons.


The reaction center (RCII) of PSII is composed of the heterodimer con-stituted by the two proteins D1 (PsbA) and D2 (PsbD), which coordinate allthe essential cofactors for its electron transport chain [9].

When the central chlorophylls special pair (P680) is excited by a photonabsorbed by the antenna apparatus, a charge separation occurs as an electronis transferred from chlorophyll to pheophytin, in one of the fastest electrontransfer reaction (2–20 ps) ever observed in nature. The cation P+

680 that isformed on this chlorophyll pair is one of the most oxidizing centers that canbe formed in a living cell. The subsequent electron transfer is to QA,, a plas-toquinone molecule tightly bound to the D2 protein near the stromal surface.This electron transfer is slower than the previous one but still fast enough(∼400 ps) to compete with recombination. This step involves a significant lossof energy that contributes to stabilize the charge separation.

Because of its very high oxidation potential, P+680 is able to extract an

electron from a tyrosine residue of the D1 protein (Tyr161, also named TyrZ)in about 20 ns; the electron hole at the donor side of PSII ends its run at theMn cluster, near the lumenal surface of the membrane. In the mean time, inthe acceptor side at the other membrane surface, the electron standing on QA

finds its way to QB , a plastoquinone loosely bound on the D1 protein, alsothanks to the assistance of a non-heme iron, sitting in between the two. Thisis the last reaction in the electron transfer chain within PSII reaction centerand its rate is the limiting step (200–300 µs).

A second turnover, driven by a further excitation from the antenna, bringsa second electron to QB with the very same reaction sequence. When QB

receives the second electron, two protons are up-taken from the stroma andthe plastoquinol leaves its binding site on D1. This is then reoccupied by anoxidized plastoquinone from the membrane pool. A two-photons-two-electronsgated mechanism brings about the reduction of a plastoquinone molecule atthe acceptor side of the complex.

In a four-photons-four-electrons gated mechanism, two more photons andtwo more electrons injected in the transport chain are required to reduce asecond PQ and oxidize two water molecules with the evolution of one oxygenmolecule. The so-called “oxygen evolving complex” (OEC) sitting in the lu-menal side of PSII, often indicated also as the “water splitting enzyme,” canbe considered as the heart of the PSII activity, and most of the today researchefforts are directed to elucidate the detailed mechanism for its activity [17].

1.4.2 Photosystem I

In photosystem I (PSI), which is located almost exclusively in stroma lamellae,the excitation of the chlorophyll special pair P700 in the reaction center ofPSI (RCI) initiates a series of electron transfer reactions that culminate inreduction of NADP+ to NADPH. RCI is activated by the absorption of redlight that promotes it to an excited state (P700*) at about −0.6 V with ajump of about 1 V with respect to the ground state P700 (E˚’= +0.4 V).


This generates a reductant strong enough to reduce ferredoxin, a soluble iron-sulfur protein able to donate electrons to NADP+ by the action of the stromalsoluble enzyme ferredoxin-NADP+ reductase. P+

700 formed is rereduced by anelectron carried by the reduced plastocyanine (Fig. 1.4). Therefore, PSI whichis situated mainly in the nonstacked, stroma lamellae regions of the thylakoidmembrane, functions as a plastocyanine–ferredoxin reductase. The core of theRCI is composed of two major subunits, PsaA and PsaB, which contain allthe pigments of the complex (including most of the antenna chlorophylls andcarotenoids) and all the electron carriers. Upon absorption of a photon, chargeseparation occurs and an electron is transferred to a primary acceptor Ao

(chlorophyll a monomer) from which it proceeds to the intermediate acceptorA1 (phylloquinone) and subsequently to the second intermediate acceptor A2

also indicated as Fx. This is a (4Fe–4S) cluster, as well as the subsequentelectron acceptors FA and FB, which transfer the electron to the ferredoxinbound in its docking site on the stromal surface of the PSI complex. P700, Ao,A1, and FX are associated with the PSI intrinsic core subunits PsaA and PsaB,while the terminal electron acceptors FA and FB are bound to an extrinsiclow molecular mass subunit PsaC [11]. It should be noticed that the electrontransfer kinetics within RCI is very fast and not completely resolved yet.

1.4.3 Cytochrome b6f

The electron transfer bridge between the two photosystems is realized bythe cytochrome b6f complex. This is assembled as a functional dimer with atwofold symmetry axis, each monomer being composed of four subunits: thecytochrome b6 carrying two b-type heme groups classified as high and lowpotential, the cytochrome f carrying a single c-type heme group, the Rieskeiron-sulphur protein carrying a (Fe2-S2) cluster and the subunit IV [13].

The reduced plastoquinoles, freely diffusing inside the lipid bilayer, bindto the complex and electrons are transferred through cytochrome b6f to theacceptor, plastocyanin. The transfer of two electrons from plastoquinol to thesingle electron carrier plastocyanin (summarized in the reaction below)

QH2 + 2plastocyanin(Cu2+) ⇀↽ Q + 2plastocyanin(Cu+) + 2H+ (1.4)

is realized through a complex series of redox reactions, which involves thesemiquinone form Q−, named “Q cycle,” with the effect of transferring 4 H+

instead of only two from the stroma to the lumen. This increases the efficiencyof converting reduction potential energy into transmembrane proton gradient.Reduced plastocyanin moves in the lumen and reaches the PSI where it willdonate its electron to P+

700 (see earlier).

1.4.4 ATP Synthase

The last complex in the photosynthetic apparatus is the ATP synthase (F-ATPase). This kind of enzyme is ubiquitous in energy-transducing membranes,


and the structure is highly conserved [12]. It presents a transmembrane do-main and a stromal portion. F-ATPase in the chloroplast synthesizes ATP,using the proton-motive force generate by H+ accumulation in thylakoid lu-men. The diffusion of protons through a channel (Fo) determines the rotationof F-ATPase inner subunits inducing cyclic conformational changes in the F1subunits, which provide the energy for ADP phosphorylation and release ofATP in the stroma; four protons are needed for the synthesis of one moleculeof ATP [18].

The equilibrium along the electron transport chain, between reduced andoxidized species, is crucial for plant life. Excess energy absorbed by chloro-phylls drives the formation of extremely dangerous reactive oxygen species(ROS), which can oxidize all the biological component of a cell. For this rea-son, absorbed energy can be redistributed among photosystems. When excesslight is absorbed by PSII, thanks to a phosphorylation process, a populationof LHCII trimers moves through the membrane and attach to PSI, thus in-creasing PSI light absorption capacity [19]. This phenomenon is called “state1–state 2 transition.”

1.5 Cyclic Phosphorylation

High electron transport rates drive the formation of reduced ferredoxin (Fd)in excess with respect to the ferredoxin-NADP+ reductase activity. To avoidover-reduction of the electron transport chain and block of proton pumpingand ATP synthesis, a mechanism, alternative to liner electron transport andinvolving PSI, can be activated, the so-called cyclic phosphorylation [20].

In this case, reduced ferredoxin can diffuse in the stroma back to thecytochrome b6f and transfer electrons to the heme exposed in this region,through the action of a Fd-NADP-reductase [21]. Translocation of the elec-tron through cytochrome to plastocyanin and then again to PSI leads to thegeneration of protons in the lumen, which can be used for ATP synthesis, evenwithout NADPH production.

This alternative pathway exists in parallel with linear electron transport;the latter is limited by the oxidation rate of PSI acceptors, thus leading, underconstant light intensity, to reach a kind of steady-state. Before the attainmentof this steady-state, if PSII is inhibited, few plastocyanin molecules move toPSI, so all the electrons that reach Ferredoxin will be used for NADP+ reduc-tion, thus inhibiting cyclic electron flow [22]. With the increase of excitationpressure, the oxidation of Fd becomes the limiting step, so the cyclic phos-phorylation is activated. In addition, the dark phase of photosynthesis is alsoinvolved in this alternative pathway, as it has been recently observed that indark-adapted plants cyclic electron transport is activated first, as a conse-quence of a scarce NADPH requirement due to Calvin cycle inhibition [23].

The two alternative electron flows can exist simultaneously because theyare structurally separated [24]. Because PSI and PSII are differentially located


within thylakoids, the position of the cytochrome b6f complex assumes a dif-ferent role: the population located near to PSII participates to linear electronflow, while the population close to PSI is involved in cyclic electron transport.This separation is driven to the extreme level in the case of C4 plants, likemaize, where PSII, thus the linear electron flow, is located in mesophyll, whilePSI, thus cyclic electron flow, is located in bundle sheath [25]. The size of thepopulations located close to PSII vs. PSI can be regulated by phosphorylationevents [26].

1.6 Photoinhibition

In describing the function of PSII, the attention has been focused on the reac-tion center, where the excitation energy coming from the antenna apparatusis converted into chemical energy through a series of redox reactions. Thecomposition of the reaction center and its electron transfer cofactors has beenbriefly discussed, but PSII core is much more complex and contains a numberof other protein subunits as always happens with enzymatic complexes thatmust be finely regulated.

In fact, a PSII core monomer contains at least 16 integral subunits, 3 lume-nal subunits, 36 chlorophylls a (two loosely attached), 7 all-trans carotenoids,2 heme groups (b and c type) belonging to the cytochrome b559, 1 non-hemeiron, 2 quinones, 2 pheophitins, and it is organized in functional dimers char-acterized by a pseudo twofold symmetry [9].

Such a great complexity is partially justified by the difficult task the PSIIcore must accomplish of splitting water bringing its electrons at the level ofplastoquinone in the membrane, but it also strongly suggests a need for afine regulation of its function. There are, indeed, several good reasons forPSII activity to be regulated: the electron transport turnover depends onthe frequency of P680 excitation, which, in turn, depends on the light fluencereaching the light harvesting apparatus. This must be large enough to ensuregood photosynthetic activity even when the environment light intensity isvery dim. But the photon fluence can vary very quickly by several ordersof magnitude. Thus, a plant, which is tuned for optimal electron transportin dim light, will find itself overexcited when exposed to full sunlight. Overexcitation of PSII brings oxidative damage to the protein-pigment complexes,thus impairing the photosynthetic activity. This is a well-known phenomenoncalled “photoinhibition” and represents the most important limiting factor tocrop productivity [27].

Therefore, a regulation is needed at the level of the excitation transferfrom the antenna to the reaction center. Indeed several sophisticated controlmechanisms have been set up by evolution for this process. Nevertheless, at alight fluence typical of midday full sun, a significant amount of photoinhibitiontakes place, which severely limits energy conversion.


Although some kind of photoinhibition has been shown to take place also atthe level of PSI [28], the main phenomenon regards PSII, where an inactivationmechanism inherent to the same electron transfer activity takes place, with lowquantum yield, even at low light irradiance [29]. At increasing light intensities,inactivation of PSII occurs with higher frequency and in extreme cases maybring to irreversible oxidative damage of the complex.

In the last decades, considerable efforts were directed to the elucidationof the molecular mechanisms of this important phenomenon. It is long knownthat the D1 protein (PsbA) of the PSII reaction center is rapidly turnedover and that its turnover rate increases with the light intensity [30]. Whyis that so? It has been hypothesized that D1 is the protein subunit, which ispreferentially damaged when the reaction center is over excited, and it needsto be frequently substituted to maintain the reaction center active. In fact, thewater splitting photochemistry of PSII produces various radicals and activeoxygen species, which cause irreversible damage to PSII. However, damagedPSII reaction centers do not usually accumulate in the thylakoid membranebecause of a rapid and efficient repair mechanism. It has been calculated thatthis repair mechanism is at least as important as that of DNA, so that thegreen kingdom in the biosphere could not survive in its absence.

It is commonly believed that the design of PSII allows protection formost of its protein and pigment components, with the oxidative damage be-ing mainly targeted to a single subunit, the reaction center D1 protein. Re-pair of PSII via turnover of the D1 protein is a complex process (Fig. 1.5)

STACKED THYLAKOID MEMBRANES

STROMA-EXPOSED THYLAKOID MEMBRANES

LIGHT

DIMERIZATIONMIGRATION MIGRATION

INACTIVATIONDAMAGE

LHC D1 D1

43 1.

7.

2.

3.

4.5.6.

3323

16 3323

164747

b559 b559

43

D2 D2 LHC

D1

43

3323

1633

2316 33

23

16

47 47 43

psbAmRNA

47 43

D2

D1

D2 D1

D1 FRAGMENTS

D1 PROTEOLYSISD1 PROCESSING

SYNTHESIS AND INSERTION OF D1LIGATION OF COFACTORS

DEPHOSPHORYLATIONOF CP43. D2 AND D1

b559 b559

b559

D2

LHC D1 D1

4333

2316 33

2316

4747

b559 b559

43

D2P P P P P P

P P

PPP

D2 LHC LHC D1

3323

16

47

b559

43

D2 LHC

PHOSPHORYLATION MONOMERIZATION

Fig. 1.5. Repair cycle of photosystem II


that involves reversible phosphorylation of PSII proteins and changes in theoligomeric structure of the complex [31]. This is combined with the shuttlingof the complex between grana and stroma-exposed thylakoid domains, partialPSII disassembly, and highly specific proteolysis of the damaged D1 protein.Replacement of degraded D1 protein with a new copy requires a complex co-ordination between its degradation, resynthesis, insertion, and assembly intothe PSII core. Although enhanced during higher irradiances, turnover of D1occurs at all light intensities, and can be easily monitored in pulse-chase ex-periments, an efficient tool to study the phenomenon. It is now clear that mostof the regulation of gene expression required for PSII repair cycle is exerted atthe translational and posttranslational levels by the redox conditions in thethylakoid membrane and in the stroma.

D1 protein together with D2 constitutes the scaffold for all the electrontransfer cofactors within RCII. It is therefore not surprising that its sacrificialhigh turnover produces profound effects on the photosynthetic activity as awhole.

There are intrinsic difficulties in understanding the rational of the pro-posed model: why should the D1 subunit be preferentially damaged by excessirradiation? How the specific proteases (belonging to the FtsH family) canactually recognize the damaged protein, considering that oxidative damageis probably a random process with no specific targets? The common viewproposes a damage-induced conformational change on the D1 protein thatwould trigger its proteolytic degradation and some experimental evidencesare brought about to support this view [32].

Let us take into consideration the main mechanism for oxidative damageinside the PSII core. When the frequency of excitation of P680 is higher thanthe rate of utilization of the reduction equivalents by PSI or, ultimately byCO2 fixing activity (Calvin-Benson cycle), an over-reduction of the plasto-quinone pool is produced and the electrons coming from RCII cannot find thephysiological exit from the complex. In these conditions, i.e., when the elec-tron acceptors are fully reduced and cannot receive further electrons, chargerecombination at P680 produces its triplet form T P680 that is able to activateoxygen molecules to their singlet state 1O2.

Normally carotenoids take care to avoid oxygen activation in the antennaby scavenging the chlorophyll triplet states by thermal dissipation. In the reac-tion center, there are actually carotene molecules, but none of them are closeenough to P680 for the chlorophyll triplet to be transferred to the caroteneand dissipated. If they were, they would immediately become oxidized by P+

680.Singlet oxygen is the main responsible of direct or indirect oxidative damageto both pigments and proteins of PSII but it is not completely clear why theaction of activated form of oxygen should be limited in their target to the D1subunit.

A possible alternative model comes from experiments on PSII fromcyanobacteria [33]. In this paper, it is shown that, under conditions of strongover-reduction of the PSII electron acceptors, massive chlorophyll bleaching


from singlet oxygen is observed in a mutant that is not able to quickly degradethe D1 protein. On the contrary, in a mutant strain in which degradation ofD1 takes place quickly, chlorophylls bleaching is very little, if any. On thebasis of these and other experimental evidences, it is proposed that triggeringof D1 degradation is not necessarily associated to an oxidative damage, whichmakes it a substrate for the proteolytic activity but, provided that the D1subunit is maintained in the right conformation, the protein is simply turnedover and replaced by the so-called “repair cycle” at a rate which is regulatedby the redox state of the chloroplast. When the excitation supply exceedsthe rate at which carbon dioxide can be fixed by the Calvin-Benson cycle,the reduction level increases and the DegP-FtsH proteases are activated todegrade D1. This mechanism ensure a protection to PSII before the oxidativedamage by activated species of oxygen takes place.

It is like the case of a prudent car driver who decides to substitute itscar tires after a given number of kilometers rather than waiting for them toblow up.

In this view, continuous degradation and resynthesis of D1 is the pricethe chloroplast must pay to preserve a very complex and delicate machinery,which is able to perform the redox chemistry that is fundamental for all formsof life on the biosphere.

References

1. A.A. Benson, M. Calvin, Ann. Rev. Plant Physiol. Plant Mol. Biol. 1, 25–42(1950)

2. H.M. Miziorko, G.H. Lorimer, Ann. Rev. Biochem. 52, 507–535 (1983)3. J. Goksoyr, Nature 214, 11614. D.J. Murphy, Biochim. Biophys. Acta. 864, 33–94 (1986)5. F. Abdallah, F. Salamini, D. Leister, Trends Plant Sci. 5, 141–142 (2000)6. J.P. Thornber, R.S. Alberte, in Encyclopedia of Plant Physiol; New Series. 5,

ed. by A. Trebst, M. Avron (Springer, Berlin Heidelberg New York, 1977) pp.574–582

7. T. Forster, Ann. Phys. Leipzig. 2, 55–75 (1948)8. G. Britton, FASEB J. 9, 1551–1558 (1995)9. K.N. Ferreira, T.M. Iverson, K. Maghlaoui, J. Barber, S. Iwata, Science 303,

1831–1838 (2004)10. Z. Liu, H. Yan, K. Wang, T. Kuang, J. Zhang, L. Gui, X. An, W. Chang, Nature

428, 287–292 (2004)11. A. Ben Shem, F. Frolow, N. Nelson, Nature 426, 630–635 (2003)12. J.P. Abrahams, A.G.W. Leslie, G.R. Lutter, J.E. Walker, Nature 370, 621–628

(1994)13. D. Stroebel, Y. Choquet, J.L. Popot, D. Picot, Nature 426, 413–418 (2003)14. S. Jansson, Trends Plant Sci. 4, 236–240 (1999)15. P.R. Chitnis, Annu. Rev. Plant Physiol. Plant Mol. Biol. 52, 593–626 (2001)16. S. Caffarri, R. Croce, L. Cattivelli, R. Bassi, Biochemistry 43, 9467–9476 (2004)17. P. Joliot, Photosynth. Res. 76, 65–72 (2003)


18. P.D. Boyer, Ann. Rev. Biochem. 66, 717–749 (1997)19. J.F. Allen, J. Bennett, K.E. Steinback, C.J. Arntzen, Nature 291, 25–29 (1981)20. D.I. Arnon, R.K. Chain, Proc Natl Acad Sci USA 72, 4961–4965 (1975)21. Z. Zhang, L. Huang, V.M. Shulmeister, Y.I. Chi, K.K. Kim, L.W. Hung,

A.R. Crofts, E.A. Berry, S.H. Kim, Nature 392, 677–684 (1998)22. P. Joliot, A. Joliot, Proc. Natl. Acad. Sci. USA 99, 10209–10214 (2002)23. P. Joliot, A. Joliot, Proc. Natl. Acad. Sci. USA 102, 4913–4918 (2005)24. P. Joliot, A. Joliot, Biochim. Biophys. Acta. 1757, 362–368 (2006)25. R. Bassi, O. Machold, D. Simpson (1985) Carlsberg Res Commun 50, 145–162

(1985)26. O. Vallon, L. Bulte, P. Dainese, J. Olive, R. Bassi, FA Wollman, Proc Natl Acad

Sci USA 88, 8262–8266 (1991)27. S.B. Powles, K.S. Chapman, F.R. Whatley, Plant Physiol 69, 371–374 (1982)28. Y. Hihara, K. Sonoike, in Advances in Photosynthesis and Respiration, vol. 11,

(Kluver Academic Publishers, Dordrecht, 2001) pp. 507–52529. B. Andersson, J. Barber, (1996) in Kluwer Academic Publishers, Dordrecht.30. J.B. Marder, A.K. Mattoo, P. Goloubinoff, M. Edelman, in Biosynthesis of the

Photosynthetic Apparatus. UCLA Symposia vol. 14, ed. by J.P. Thornber, L.A.Staehelin, R.B. Hallick (Liss, A.R., New York, 1984) pp. 309–311

31. R. Barbato, G. Friso, F. Rigoni, F.D. Vecchia, G.M. Giacometti, J. Cell Biol.119, 325–335 (1992)

32. E-M Aro, T. Hundal, I. Carlberg, B. Andersson, Biochim. Biophys. Acta. 1019,269–275 (1990)

33. E. Bergantino, A. Brunetta, E. Touloupakis, A. Segalla, I. Szabo, G.M.Giacometti, J. Biol. Chem. 278, 41820–41829 (2003)

2

Exploiting Photosynthesis for BiofuelProduction

C. Govoni, T. Morosinotto, G. Giuliano, and R. Bassi

During the recent “Energy Outlook and Modeling Conference” of March 2006in Washington DC, it was estimated the world current energy consumption in100 “Quads,” where one Quad corresponds to about 25 million of oil equiv-alent tons (MTep, http://www.eia.doe.gov/oiaf/aeo/conf/). Nowadays, over85% of world energy demands are met by the combustion of fossil fuels: coal,oil, and natural gas. Current oil reserves are estimated to be about 1,277billions of barrels and, assuming a stable consumption, they would be suf-ficient for next 42 years. However, in the middle-term oil utilization is ex-pected to rise of 1.6% every year thus making reserves exhaustion even faster(http://www.eia.doe.gov/oiaf/aeo/conf/pdf/petak.pdf). It is thus clear thatto sustain our lifestyle to find alternative and renewable sources of energy isa striking and urgent need.

Besides the problem of fossil reserves depletion, the massive combustionof fossil fuels in the past decades also had a high environmental impact. Infact, this leads to releases of large amounts of carbon dioxide and other pollu-tants in the atmosphere. However, these emissions are reassimilated by naturalprocesses, which cannot keep the pace of the present CO2 production rates.In fact, every year the forests are able to fix about 1 billion of tons of carbonin organic matter and further 2 billion of tons are fixed in the ocean everyyear by the sea photosynthesis, but the CO2 emissions caused by the humanactivity are about 6 billion of tons. So the balance is positive and every year3 billion of tons increase the global level of CO2 in the atmosphere (Fig. 2.1).

The alteration of this equilibrium caused an accumulation of carbon diox-ide in the atmosphere, which is estimated to have risen of about 15% since1750. The main consequence of the accumulation of CO2 in the atmospherehas been proposed to be the “greenhouse effect” some gases like carbon diox-ide, in fact, are able to retain the infrared radiation into the atmosphere andcause a global temperature increase of the planet. The correlation betweenCO2 and temperature increase is clear, although a cause/effect relationshiphas not been clearly established, yet.

16 C. Govoni et al.

Fossil FuelsRelease: 6 Net Ocean

Uptake: 2

RespirationMicrobial

Decomposition

Atmospheric net carbon Increase 3-4 Gt / y

SedimentsRocks

Soil

Plant Biomass

Deep ocean

Net TerrestrialUptake: 0-1

Photosynthesis

Photosynthesis

Fig. 2.1. Simplified global carbon cycle. Biogeothermal carbon cycle is schematized,distinguishing between terrestrial and aquatic processes. Photosynthesis, which fixesatmospheric CO2 is indicated in black, while processes responsible of carbon dioxiderelease (respiration, microbial decomposition, and fossil fuels combustion) are indi-cated in italic. The net uptake activity of terrestrial and ocean activities are circled.Plants biomass, soil, rock, deep ocean, and marine sediments are indicated in whiteand are sites of long-term carbon storage.

This temperature increase has two major consequences. The first is deser-tification of subtropical areas, which caused a massive reduction of forests. Astrees are photosynthetic organisms, which fix the atmospheric carbon dioxideinto biomass, their massive reduction is causing further enhancement of car-bon dioxide accumulation. Although a transitory effect, even more importantis that desertification implies oxidation to CO2 of the organic mass containedin the soil (1–2%). The second effect consists into the thawing out of the polarareas. In these regions, a large amount of CO2 has been fixed as frozen organicmatter in the underground during several million years. Because of the globalwarming, this organic matter is being thawed out and mineralized by bacteriaand fungi causing a further CO2 evolution.

The public conscience that the search of energy sources alternative to fossilfuels is fundamental for future prospects is generally increasing, and researchefforts are devoted to this field in several countries. The attention is focusedon renewable resources, such as sun and wind, which can be considered as noexhaustible. Furthermore, their massive exploitation will drastically reducechemical, radioactive, and thermal pollution, and they therefore stand out asa viable source of clean and limitless energy.

2 Exploiting Photosynthesis for Biofuel Production 17

Among the possible renewable energy sources, an interesting emergingoption is the exploitation of higher plants or algae and their capacity to harvestsolar light and convert it into chemical energy. This option is particularlyinteresting over other methods of solar radiation exploitation because theseorganisms are able to convert CO2 into biomass, through the photosynthesis,and thus their utilization will also contribute to reduce carbon dioxide levels.

A further advantage is that, unlike fossil fuels, photosynthetic organismsare rather uniformly distributed over earth surface. Thus, their exploitationwould not require large plants, and energy will be produced locally in smallbio-power plants. This different spatial distribution of power plants has alonea positive effect on energy efficiencies: in fact, as a relevant percentage ofenergy is lost only during its transport from production to utilization, thereduction in distance would reduce significantly energy production needs.

Possible applications of plants or algae for energy production are described,with particular attention on their possible exploitation for biofuels and bio-hydrogen production.

2.1 Biological Production of Vehicle Traction Fuels:Bioethanol and Biodiesel

Any fuel derived recently from living organisms or from their metabolic by-product is defined as a biofuel. Unlike other natural resources like oil, coal,and nuclear fuels, biofuels are a renewable energy source. They present theadvantage of a cleaner combustion in comparison to conventional fuels, andthus have a lower impact on atmospheric CO2 levels.

Liquid biofuels are mainly developed for vehicle traction. At present themost promising options are bioethanol and biodiesel.

2.1.1 Bioethanol

Bioethanol can be utilized as a fuel in combustion engines in different ways:(1) as hydrous ethanol (95% by volume), containing small percentage of water,it can be used as a gasoline substitute in cars with modified engines. (2) Asanhydrous (or dehydrated) ethanol, free from water or at least 99% pure, itcan be blended with conventional fuels in ratios between 5 and 85% (E85).As a 5% component, it can be used in all recently produced engines withoutmodification. Higher blends require modified engines installed in the so-calledflexible fuel vehicles. (3) Finally, bioethanol is also used to manufacture ethyl-tertiary-butyl-ether (ETBE), a fuel additive for conventional gasoline.

Bioethanol is produced from starch plants (like grain, corn, and tubers-like cassava), sugar plants (sugar beet or sugar cane), and – although not yeton a large scale – from cellulose. Carbohydrates present in these plants areused as a substrate for the microbial fermentation, and ethanol is subsequently

18 C. Govoni et al.

enriched by distillation/rectification and dehydration. The CO2 liberated dur-ing its combustion was first fixed from the atmosphere by the crops duringtheir growth, and thus the use of bioethanol in place/addition of fossil fuelswould then reduce the net release of greenhouse gases in the atmosphere. Thepotential impact of this strategy is large if we consider that road transport ac-counts for 22% of all greenhouse gas emissions (http://:www.foodfen.org.uk).Bioethanol would have a positive environmental impact also because it isbiodegradable and not toxic for living organisms as fossil fuels. A furtheradvantage is that bioethanol can be easily integrated into the existing fueldistribution system. In fact, as mentioned, in quantities up to 5%, bioethanolcan be blended with conventional fuel without the need of engine modifica-tions and adaptation of forecourts and of transportation system.

Finally, its increased use would also have positive a socio-political impact,by reducing the dependence from the oil producing countries and boostingthe rural economy.

2.1.2 Biodiesel

Biodiesel is an alternative fuel, which can be produced from vegetal resources.It is made by the trans-esterification of vegetal lipids generating lipid methylesters (the chemical name for biodiesel) and glycerine, which is a valuableby-product used in soaps and other products.

In USA, biodiesel is mainly produced from soybean oil, while in Europefrom rapeseed oil. In addition to these crops, biodiesel can be produced usingsome microalgae, which are able to accumulate large amount of triglyceridesin their cells. Biodiesel contains no petroleum oil, but it can be blended atany level with petroleum diesel, and in this form it can be used in recentcompression-ignition (diesel) engines with little or no modifications. Thus itsintroduction, even if massive, would not need modification in the fuel dis-tribution systems. It also shares other advantages with bioethanol. As it isproduced from vegetal sources, its utilization would have no net impact onthe concentration of carbon dioxide in the atmosphere. Furthermore, biodieselis also biodegradable, nontoxic, and essentially free of sulphur and aromatics.

2.1.3 Biofuels Still Present Limitations Preventing Their MassiveUtilization

Although bioethanol and biodiesel present big advantages for the environmentand they could rather easily substitute at least partially fossil fuels, there arestill two major limitations to the massive use of biofuels: energetic yield andcosts.

First, the energetic yield is calculated as the ratio between all the energeticinputs (used for the cultivation of the crop, transport, and transformation)and all the outputs (fuel, but also all the by-products, as glycerine in thecase of biodiesel). The energetic yield of biodiesel from soya, for example, is


about 3.2, meaning that for every unit of energy employed in the productionof biodiesel, 3.2 units are harvested at the end of the process. Performanceswith bioethanol from maize are even less encouraging being its energetic yieldabout 1 or even less [1, 2].

Second, considering the present oil price, biofuels are still not economicallycompetitive. In fact, biodiesel is estimated to be economically convenient overthe limit of 80$/barrel (and even higher for bioethanol). So, at least on a shortterm, biofuels introduction in the commercial networks needs to be financiallysupported by governments willing to stimulate their utilization. However, ina medium timespan perspective, when oil reserves will be exhausting and theoil price will consequently rise, biodiesel will be a major option for vehicletraction and investments will be paid back.

2.2 Hydrogen Biological Production by FermentativeProcesses

Hydrogen has some very interesting properties, which makes it the potentialenergy source of the future. It is easily converted into electricity, and it hasvery high energy content per weight unit. Moreover, its utilization has thelowest possible environmental impact. It produces electricity by reacting withoxygen and yielding water as the only waste. However, these advantages areeffective only if hydrogen is produced by an efficient and clean process. Un-fortunately, this is currently not the case, and hydrogen is mainly derivedfrom fossil fuels and its production generates considerable amounts of CO2 inaddition to other air pollutants such as sulphur dioxide and nitrogen oxides(http://www.fao.org/docrep/w7241e/w7241e05.htm).

Thus, hydrogen’s positive impact on the energy system depends on thedevelopment of new methods for its efficient production. One major option isthe water electrolysis, which, however, requires an efficient, clean, and cheapmethod for electricity generation, which is at present not available.

For this reason, other options are receiving an increasing attention and,among them, the biological H2 production. In fact, some living organismsare able to produce H2 in anaerobic conditions thanks to metabolic reactionscoupled with fermentation, nitrogen fixation, or photosynthesis. This is a verypromising approach, as it has the potential to provide cheap hydrogen withoutusing fossil fuels and emitting pollutants.

Living organisms can synthesize hydrogen from protons and electrons, us-ing special enzymes named hydrogenases. Hydrogenases fall into two majorclasses: Ni-Fe and Fe-Fe hydrogenase identified from the metal content intheir active site [3]. They differ in the composition and in the structure ofthe metal center active site as well as in their polypeptide sequence, but alsoshow some conserved properties, namely the presence of CN and/or CO lig-ands in the active site. In general, Fe-Fe hydrogenases were shown to be more

20 C. Govoni et al.

efficient catalysers, and therefore they are more promising for practical ex-ploitations [4]. Both Fe-Fe and Ni-Fe hydrogenases have a strong sensitivityto O2 presence, probably because of its binding to the metallocatalytic sites.For this reason, all hydrogen producing organisms identified till very recentlyare active in anaerobic conditions only. Exceptions to this pattern are stillunder evaluation.

Several organisms can produce hydrogen and differ for the source of reduc-ing power they use for this biosynthesis. Some bacterial strains can exploitorganic compounds by fermentation, while other organisms, such as greenalgae, are able to use directly the solar radiation harvested by the photo-synthetic apparatus. However, we must be aware that, despite these strongpotential advantages, several bottlenecks need to be resolved before practi-cal application can be economically sustainable and further research is stillneeded.

2.2.1 Hydrogen Production by Bacterial Fermentation

Several bacteria are able to produce hydrogen from the fermentation of car-bohydrates in anaerobic conditions. As substrates for this reaction, severalsources could be used. However, the use of human and industrial wastes isparticularly interesting as it attains two objectives at the same time: process-ing the waste and produce environmental-friendly energy.

Anaerobic bacteria use organic substances as source of electrons and en-ergy, converting them into hydrogen. The reactions involved in hydrogen pro-duction (2.1 and 2.2) are rapid and suitable for treating large quantities ofwastewater in large scale fermenters.

Glucose + 2H2O = 2Acetate + 2CO2 + 4H2 ∆G = −184.2 kJ (2.1)

Glucose = Butyrate + 2CO2 + 2H2 ∆G = −257.1 kJ (2.2)

The produced molecules, acetate and butyrate, still have a residual chemicalenergy, which remains unexploited. This is because the complete oxidation ofcarbohydrates to H2 and CO2 is not a spontaneous process as it has a ∆Gof only −46 kJ mol−1 hydrogen. There are no known fermentation pathwaysthat can achieve a conversion efficiency larger than 4 mol H2/mol hexose [5].

To achieve a further decomposition of the remaining organic substances, anexternal energy input is necessary. This can be derived, as recently described,from the application of an external small electric potential, which stimulatesthe fermentation of acetate to H2 in modified microbial fuel cells [6].

Another valuable option is to couple the first fermentation step by anaero-bic bacteria with the second one performed by photosynthetic sulphur bacte-ria, which use the residual organic acids to produce hydrogen (Fig. 2.2). Thislast step, which completes the fermentation of organic wastes to hydrogen,can be performed in these organisms, even if energetically unfavorable, asthey employ light energy to drive the reaction.


LIGHT-DRIVENFERMENTATION

BIOMASSCARBOHYDRATESFERMENTATION

H2 + CO2

H2

CO2

H2 + CO2

Organic Acids

LIGHT

Gas Separator

Fig. 2.2. Hydrogen production by dark and light fermentation of organic wastes.Carbohydrates from several sources such as human or industrial wastes are fermentedfrom anaerobic bacteria into organic acids, carbon dioxide, and hydrogen. The formerare decomposed by photosynthetic bacteria, using light energy to drive the reaction.Gases yields from both fermentation processes are separated by a gas separator,yielding pure hydrogen.

Clearly, more research and development is required but, when developed,H2 fermentations from organic wastes would be relatively low cost comparableto methane fermentation (in the range of $4–8/MBTU1) as they could also usesimilar mixed-tank reactors, without need for the building of new installations.Thus, dark H2 fermentation of wastes is close to be competitive with fossil fuel-derived H2, providing a first approach to large scale biohydrogen production.

2.3 Hydrogen Production by Photosynthetic Organisms

In addition to the mentioned sulphur bacteria, other photosynthetic microor-ganisms are able to produce hydrogen. However, they do not use light energyonly to drive a fermentative reaction, but also to directly decompose waterto hydrogen and oxygen. This possibility is clearly very attractive from the

1 British thermal unit. One Btu is equal to the amount of heat required to raisethe temperature of one pound of liquid water by 1◦F at its maximum density,which occurs at a temperature of 39.1◦F. One Btu is equal to ∼251.9 calories or1,055 J.

22 C. Govoni et al.

point of view of environmental impact because this would be the perfect wayto produce energy, using only two renewable resources, solar radiation andwater.

The photosynthetic microorganisms able to produce hydrogen arecyanobacteria and green algae.

2.3.1 Cyanobacteria

Cyanobacteria can use two different enzymes to generate hydrogen gas [7].The first is nitrogenase, which catalyzes the fixation of nitrogen into ammo-nia, producing hydrogen as a by-product. Hydrogen photo-evolution catalyzedby nitrogenases requires anaerobic conditions, as hydrogenases. As oxygen isa by-product of photosynthesis, cyanobacteria have developed two differentstrategies to achieve anaerobiosis:

1. Some evolved the capacity of building structures with reduced oxygen-permeability, called heterocysts, thus separating physically oxygen evolu-tion from nitrogenase activity [8].

2. Nonheterocystous cyanobacteria instead separate temporally oxygen evo-lution from nitrogenase activity activating the latter only during darkperiods.

Whatever is the case, hydrogen production by nitrogenase enzyme is in-teresting but presents an important limitation. As the reaction is coupled tothe nitrogen fixation, it is energetically demanding, making the whole processpoorly efficient.

The other hydrogen-metabolizing enzymes in cyanobacteria belong to theclass of hydrogenases. One class, uptake hydrogenase (encoded by hupSLgene), however, catalyzes the oxidation of hydrogen molecules to produce pro-tons and electrons [9]. Uptake hydrogenase enzymes are found in the thylakoidmembrane of heterocystis from filamentous cyanobacteria, and they transferthe electrons from hydrogen for to the respiratory chain. Nitrogenases of thistype are not responsible for net hydrogen production in vivo, rather, allow forpartial recuperation of energy invested in N2 fixation and employed for hydro-gen production. In the perspective of the production of hydrogen molecules,their activity is not desirable and should be inhibited.

The second type of hydrogenases are reversible, or bi-directional hydroge-nases (encoded by hoxFUYH ) that can either consume or produce hydrogen.The biological role of these bidirectional or reversible hydrogenases is poorlyunderstood and thought to be involved in the ions level control of the organ-ism. Reversible hydrogenases are associated with the cytoplasmic membraneand likely function as electron acceptor from both NADH and H2 [10]. If op-portunely regulated, these enzymes can be exploited for hydrogen production.


2.3.2 Eukaryotic Algae

Gaffron and coworkers over 60 years ago discovered that green algae, whenilluminated after an anaerobic incubation in the dark, have the ability toproduce H2 [11, 12]. This hydrogen production is directly coupled with wa-ter oxidation through Photosystem II (PSII) and the photosynthetic electrontransport chain. Thus, H2 is synthesized directly from water, using sunlightas a source of energy, with a totally clean process, making very promising theperspective of exploiting green algae for a large scale H2 production. More-over, differently from the cyanobacteria, H2 production is achieved by Fe-Fehydrogenases, thus in a more efficient way.

2.4 Challenges in Algal Hydrogen Production

The exploitation of algae for hydrogen production is probably the method withthe best perspectives in the long term among those that are analyzed here.However, it has two major limitations that prevent its large scale exploitation:(1) the great sensibility of the process to the oxygen presence; (2) the inefficientlight energy distribution in the bioreactor.

2.4.1 Oxygen Sensitivity of Hydrogen Production

As mentioned, hydrogen is always produced in anaerobic conditions becauseall hydrogenases are sensitive to oxygen. This limitation becomes deleteriouswhen oxygenic photosynthetic organisms such as algae are discussed: in fact,they use water as electron donor to produce protons and oxygen. The latter isthus an unavoidable by-product of photosynthesis and even the incubation inanaerobic conditions is sufficient to avoid inhibition of hydrogen production.

Different strategies are possible to obtain a significant level of hydrogenproduction by photosynthetic organisms: one is to separate temporally or spa-tially oxygen evolving process from the hydrogen production. Alternatively,modified hydrogenase enzymes, less sensitive to O2, can be searched.

Some cyanobacteria, as mentioned, can maintain a state of controlledanaerobiosis in heterocystis; however, it is difficult to imagine modifying greenalgae to make them able to build such complex structures. In the case of algae,thus, the approach of a temporal separation of oxygen and hydrogen evolutionsis more promising. In 2000, Melis et al. [13] demonstrated in Chlamydomonasreinhardtii that, when sulphur in the growth medium is limiting, rate of oxy-genic photosynthesis declines without affecting significantly the mitochondrialrespiration. In fact, under illumination Photosystem II (PSII) polypeptides,in particular the D1 subunit, have a very fast turnover. Under sulphur de-privation, when the protein biosynthesis is inhibited, PSII is thus among thefirst protein complexes affected, causing the decrease of photosynthetic ac-tivity and oxygen evolution. On the contrary, the mitochondrial complexes,

24 C. Govoni et al.

which have a lower turnover rate, can maintain longer their function. In sealedcultures, the imbalance between photosynthesis and respiration results in anet consumption of oxygen by the cells and the establishment of anaerobiosis,which stimulates the H2 production.

Sulphur deprivation, however, is a stress condition for the cells and after afew hours H2 production yield decreases. Therefore, it is necessary to restore anormal level of sulphur in the culture to allow the cells to recover, re-establishphotosynthesis and reconstitute energy reserves.

However, the process is reversible and after a recovery delay the sulphurdeprivation treatment can be repeated for further induction of hydrogen pro-duction. In fact, it has been demonstrated that phases of S deprivation andH2 production can be alternated to phases where sulphur is supplied and H2

is not produced. The maximum energy yield obtained by this method is ap-proximately 10%, so close to the best photovoltaic cells. The cost is lower,however, as solar cells need to be built while algal cells reproduce. However,the biological systems can reach maximum yields only in a transitory way,while in the long-term average yields are still too low (0.1–1%).

In addition to the still insufficient yields, the implementation of thismethod of hydrogen production on a large scale system is complicated bythe need of changing the growing medium to add or remove sulphur. For thisreason, it is important to search more efficient methods to induce anaerobio-sis, as PSII is the only responsible of oxygen evolution, and the regulation ofits expression is a relatively simple method to induce or suppress anaerobiosis.For this reason in the perspective of a large scale application, it would be morepractical to have Chlamydomonas strains where one or more genes essentialfor the PSII assembly are under the control of an inducible and easily con-trolled promoter. If the presence or absence of PSII can be readily controlledin the culture, it would be possible to control the succession of anaerobiosisand aerobiosis.

The oxygen sensitivity of the hydrogen production is mainly due to thesensitivity of the enzyme hydrogenase itself. In fact, these enzymes are inacti-vated by the presence of very low oxygen concentrations. The exact reason forthis sensitivity is not fully understood at molecular detail and is the subjectof intense research. However, as different hydrogenases have variable struc-tural organization but share a similar active site, the nature of this lattercomponent is probably causing oxygen sensitivity.

The most efficient method to obtain large scales hydrogen productionwould be to find an oxygen resistant hydrogenase. To this aim two differ-ent approaches are possible: first, to search for hydrogenase mutants with anhigher selectivity for hydrogen. In particular, the idea of reducing the oxy-gen diffusion to the active site by modulating the molecular size of the gaschannels within the molecule looks promising. The second possible approachis to find organisms that can synthesize hydrogen even in presence of oxygenand thus naturally have oxygen resistant hydrogenases. In this respect, it isinteresting the finding that such organisms may actually exist in nature, as


demonstrated by the example of Thermotoga neapolitana, which was shown toproduce hydrogen even in the presence of 4–6% oxygen [14]. The gene encod-ing the hydrogenase of this organism has been isolated and sequence analysissuggested that its superior resistance could be due the fact that this enzymeis generally found as a trimer, while all the other hydrogenases known aremonomers. Thanks to this larger molecular size, so that its active site canbe buried and less accessible to oxygen molecules. However, this explanationalone is insufficient as a homologous organism, T. maritima has an hydro-genase with similar size but it is not equally resistant to oxygen. In a closefuture perspective, the study of this hydrogenase would suggest possible waysto further improve oxygen resistance, while the expression of this resistanthydrogenase in Chlamydomonas cells would allow understanding the possibleimpact of a oxygen-resistant hydrogenase on hydrogen yields.

2.4.2 Optimization of Light Harvesting in Bioreactors

The first step in hydrogen production is light absorption by photosyntheticcomplexes. When algae grow in photobioreactors, this process has a poor effi-ciency because of the bad distribution of light. In fact, to have good produc-tion yields, biomass concentration need to be high, leading to increased opticaldensity of the culture where light can only penetrate for a few millimeters,as shown in Fig. 2.3. As a consequence, the external cell layers are exposedto a very intense light leading to activation of photo-protective mechanismsto dissipate energy in excess, to avoid formation of oxygen reactive speciesand consequent cellular damages. This cells population, thus, is not efficientin the light utilization because absorbed energy is dissipated rather than usedfor metabolic reactions. On the contrary, cells located at more internal levelin the culture are exposed to very low light, insufficient for sustaining intensemetabolic activity.

This unequal light distribution has another negative consequence: the con-nective fluxes in the medium cause algal cells to stir into the reactor wherethey can move from dark to strong light within few seconds. The fast variationin light intensity leads to an input of reducing power into the photosyntheticchain at very different rates, without allowing the time for acclimation ofphotosynthetic apparatus [15]. These conditions are highly stressful because,without the delay needed for the activation of photo-protective mechanisms,the energy absorbed in excess generates oxidative stress and leads to PSIIphotoinhibition.

To overcome these problems and increase the productivity of photobiore-actors, two type of interventions are needed:

1. To reduce the antenna size to lower the optical density of the culture andallow for a better light distribution within the bioreactor without reducingthe cellular density

2. To strengthen the algal mechanisms of oxidative stress resistance and ofthermal energy dissipation

26 C. Govoni et al.

MetabolicMetabolicactivityactivity

Light Intensity

Heat Heat DissipationDissipation

Heat Heat DissipationDissipation

WT strain

Strain with reduced antenna

MetabolicMetabolicactivityactivity

IncreasingDepth

IncreasingDepth

Light Intensity

Fig. 2.3. The limits of light distribution in photobioreactors yields. In photobiore-actors, a high cell concentration should be maintained to obtain a reasonable yield.This leads to unequal light distribution. In fact, more exposed cellular layers (left)absorb a very high light, which can be used only partially for metabolism. A largefraction is also dissipated thermally to avoid establishing oxidative stress in cells.Internal layers (right) are illuminated by a very poor light, insufficient to supportintense metabolic activity. If a strain with reduced antenna is used in the photo-bioreactor, with an equal cellular concentration, the optical density of the culture islower and light is more efficiently distributed. External cells absorb a lower propor-tion of available light, which thus reaches more internal layers supporting metabolicactivity.

Among the different mechanisms, it is important to increase the capacity torespond to fast alterations in light intensity.

The only strategy attempted so far to increase the light distribution in pho-tobioreactors has been the use of mutants lacking all the antenna proteins, aclass of proteins that are responsible of a large fraction of light harvesting inphotosynthetic eukaryotes. With this approach, light distribution has been in-deed improved. However, cells suffered for strong photo-sensitivity and did notsurvive when exposed to full sunlight. This puzzling phenotype ( lower antennasize is expected to decrease the number of photons funnelled to RCII-reaction


centre II- and thus reduce sensitivity) is explained by the observation thatantenna proteins (named Lhc) are not only involved in light harvesting butalso have a major role in photoprotection [16]. As a consequence, the deletionof all antenna proteins reduces the capacity for photoprotection to a levelmaking insignificant the advantage obtained in terms of light distribution.

However, among Lhc polypeptides, individual members proteins are spe-cialized in light harvesting, while others in photoprotection [17]. Thus, toobtain mutants optimized to bioreactor conditions, the first step consists inidentifying members of Chlamydomonas Lhc protein family essential for pho-toprotection vs. those that can be deleted to lower the optical density.

To this aim, we apply two different approaches: in vivo approach consistsin the creation of a library of Chlamydomonas insertional mutants. In thesemutants genes are randomly knocked out by the insertion of a resistancecassette. These mutants are screened for a smaller antenna size, by analyzingtheir fluorescence yield properties and Chl a/b ratio. Strains with a reducedantenna content, in fact, would have lower cell fluorescence yield and higherChl a/b ratio, as Chl b is specifically bound to Lhc proteins.

An in vitro approach is complementary: individual members of Lhc fam-ily are studied one by one. Each Lhc protein identified in the genome [18]is characterized by expressing in bacteria the corresponding polypeptidesand refolded in vitro to obtain the native pigment protein complex. By thismethod, each protein will be characterized individually for its capacity of lightharvesting and photoprotection.

References

1. A.E. Farrell, R.J. Plevin, B.T. Turner, A.D. Jones, M. O’Hare, D.M. Kammen,Science 311, 506–508 (2006)

2. C.N. Hunter, J.D. Pennoyer, J.N. Sturgis, D. Farrelly, R.A. Niederman, Bio-chemistry 27, 3459–3467 (1988)

3. P.M. Vignais, B. Billoud, J. Meyer, FEMS Microbiol. Rev. 25, 455–501 (2001)4. M.W. Adams, Biochim. Biophys. Acta 1020, 115–145 (1990)5. R.K. Thauer, K. Jungmann, K. Decker, Bacteriol. Rev. 41, 100–180 (1977)6. H. Liu, S. Grot, B.E. Logan, Environ. Sci Technol. 39, 4317–4320 (2005)7. D. Dutta, D. De, S. Chaudhuri, S.K. Bhattacharya, Microb. Cell Fact. 4, 36

(2005)8. B. Bergman, J.R. Gallon, A.N. Rai, L.J. Stal, FEMS Microbiol. Reviews 19,

139–185 (1997)9. P. Tamagnini, E. Leitao, F. Oxelfelt, Biochem. Soc. Trans. 33, 67–69 (2005)

10. G. Boison, H. Bothe, A. Hansel, P. Lindblad, FEMS Microbiol. Letters 174,159–165 (1999)

11. H. Gaffron, Nature 204–205 (1939)12. H. Gaffron, J. Rubin, J. Gen. Physiol. 26, 219–240 (1942)13. A. Melis, L. Zhang, M. Forestier, M.L. Ghirardi, M. Seibert, Plant Physiol. 122,

127–136 (2000)

28 C. Govoni et al.

14. S.A. Van Ooteghem, A. Jones, L.D. Van Der, B. Dong, D. Mahajan, Biotechnol.Lett. 26, 1223–1232 (2004)

15. A.V. Vener, P.J. van Kan, P.R. Rich, I.I. Ohad, B. Andersson, Proc. Natl. Acad.Sci. USA 94, 1585–1590 (1997)

16. D. Elrad, K.K. Niyogi, A.R. Grossman, Plant Cell 14, 1801–1816 (2002)17. P. Horton, A. Ruban, J. Exp. Bot. 56, 365–373 (2005)18. D. Elrad, A.R. Grossman, Curr. Genet. 45, 61–75 (2004)

3

In Between Photosynthesisand Photoinhibition: The Fundamental Roleof Carotenoids and Carotenoid-BindingProteins in Photoprotection

G. Bonente, L. Dall’Osto, and R. Bassi

3.1 When Light Becomes Dangerousfor a Photosynthetic Organism

During operation of oxygenic photosynthesis, highly reactive molecules suchas excited chlorophylls are placed in an environment, the chloroplast. This isprobably the very spot where oxygen concentration is the highest on earth.There are two major sources of reactive oxygen species (ROS) in the pho-tosynthetic apparatus. The first source is electron transport, where electronsare extracted from water and transported to NADP+. The powerhouses ofphotosynthetic electron transport are the PSI and PSII reaction centers, whichtransfer one electron at time. Now, relatively stable oxygen forms are, respec-tively, the most oxidized one (O2) and the most reduced one (H2O). All theother intermediate redox states are highly reactive and are, in fact, calledROS. It can easily be understood that during a multistep electron transportchain, the occasions in which O2 can be exposed to reduction by a singleelectron thus yielding superoxide are easily produced. Moreover, in PSII, thesite where electrons are extracted one by one from water, intermediate redoxstates may be produced.

The second major source of ROS is the process of light energy absorp-tion by chlorophylls and transfer of the excitation energy between the manychlorophylls forming the antenna system. Plants are particularly prone tophoto-oxidative damage, and for the same reasons they are effective at photo-synthesis, namely because the primary pigment, chlorophyll (Chl) is a veryefficient sensitizer. In fact, chlorophyll has a long living singlet state (5 ns) thusallowing for intersystem crossing and formation of triplet chlorophyll-excitedstates, which can react with O2, a triplet in its ground state, to yield 1O∗

2 .This reactive species as well as others deriving from its reaction with waterand organic molecules cause oxidative damage to occur in proteins, lipid,and pigments, leading to photoinhibition of photosynthesis and, ultimately,to photobleaching of pigments.

30 G. Bonente et al.

1Chl*1Chl*

1Chl

3Chl*TPathway 2:Fluorescence

Pathway 1: Photochemistry

Reaction CenterPathway 3:Heat (NPQ)

Pathway 4:Intersystem crossing

S1

S0

Fig. 3.1. The singlet chlorophyll (S1) deexcitation pathways: Pathway 1, energyutilization in photosynthesis (photochemical quenching); pathway 2, the energyreemitted as fluorescence; pathway 3, thermal dissipation pathway (NPQ, nonpho-tochemical quenching); pathway 4, the singlet chlorophyll conversion into triplet(intersystem crossing)

When does this happens and how? When a photon is absorbed by a chloro-phyll molecule, the molecule is promoted to its first singlet excited state. Thisexcited state does not live long because it is usually dissipated through dif-ferent pathways (Fig. 3.1); the most important (for photosynthesis) is photo-chemistry, i.e., charge separation and electron transport. If this pathway isactive, most of the energy is used through this way and Chl lifetime is below1 ns. The second pathway is fluorescence: the photon is reemitted and Chl goesback to its ground state. This pathway is less important and fluorescence isusually very low (below 1%); however, it is very useful for scientists as it tellsus in any moment the level of excited states in the photosynthetic system. Thethird path is heat dissipation, which can be extremely variable. Normally, itchanges in a complementary way with respect to photochemistry. When thephotochemistry is high, there is very little heat dissipation and the yield of thephotosynthetic process is high. When some factor (water, temperature, CO2)limits photochemistry, then mechanisms are activated for unharmful dissipa-tion of the excitation energy into heat. The reason for this careful control isactually to keep as low as possible the number of excitons entering the lastpathway of deexcitation: intersystem crossing (IC). Intersystem crossing isthe conversion, through electronic spin inversion, of chlorophyll single-excitedstate into triplet-excited state.

As light increases, chlorophyll-excited states concentration stays low asfar as photochemistry keeps the pace. When photochemistry is limited, heatdissipation is activated, after saturation of heat dissipation pathway, no fur-ther prevention is available and the probability of intersystem crossing occurincreases, followed by photodamage. Photodamage can be limited by anotherset of mechanisms collectively indicated as “ROS scavenging.” Singlet oxygenseems to be the main species produced in photoinhibited membranes, otherreactive species are involved, as superoxide anion (O−

2 ), which is produced atboth PSII oxygen evolving complex and PSI (Mehler reaction), hydrogen per-oxide (H2O2), and hydroxyl radical (.OH), species which are easily produced

3 Carotenoids Proteins in Photoprotection 31

when the electron transport carriers are over-reduced. A major target is mem-brane lipids, which undergo peroxidation. This proceeds by a free radical chainreaction mechanism. The level of lipid peroxidation, in fact, is one of the mostvaluable indicators for photooxidative damage.

This article reviews several photoprotective processes that occur withinchloroplasts of eukaryotic photosynthetic organisms, starting with mediumand long-term response strategies to photooxidative stress, as state transi-tion and acclimation. Following, a particular emphasis will be laid on themolecular mechanisms, which prevents ROS formation or participate in ROSdetoxification inside the chloroplast. In these mechanisms, carotenoids andcarotenoids-binding proteins, which belong to the Lhc family, have a key role.

3.2 Acclimation

When photosynthetic organisms are exposed to different light conditions withrespect to those of growth, acclimation processes are activated, which consistinto remodeling of the PSII composition. In case of acclimation to high light,e.g., we observe a strong decrease of LHCII trimeric complexes yielding a lowercross section for each PSII reaction center complex, while PSI complexes actas a pivot reference and do not change their composition [1]. An extremeexample of such acclimation has been described in the green alga Dunaliellasalina, where the chlorophyll antenna size of PSII has been reported to be assmall as 60 chlorophyll molecules under high light conditions, while as largeas 460 chlorophyll molecules under low light growth [2]. The light-harvestingantenna reduction seems to be a common strategy, across evolutionary distantorganisms, adopted to optimize photosynthesis and avoid ROS production instressing conditions. Besides the ratio between antenna and reaction centercomplexes, the stoichiometry of electron transport components with respect tolight-harvesting components also changes to sustain an higher electron trans-port rate. Also, enzymes involved in metabolic sinks, such as the Calvin cycle,increase their relative abundance and activity [1]. The mechanism for stoichio-metric adaptation of PSII antenna size is posttranslational: accumulation ofzeaxanthin induced by the excess light [3] binds to Lhc proteins into a spe-cific allosteric site called L2 [4], thus inducing a conformational change [5, 6],which leads to monomerization and degradation of LHCII but not of minorLhcb components [7]. Increase in the stoichiometry of others electron trans-port components, such as cytochrome b6f complex, is likely to be controlledby transcriptional activation [8].

Although acclimation to constant light conditions has been studied tosome extent, the resulting architecture of the photosynthetic apparatus canonly reach a compromise condition reflecting the average conditions over aperiod of several days, as shown by the need of at least three days excess lightperiod to elicit adaptative proteolysis of LHCII [9].


In fact, over a single day, light intensity and temperature change dramati-cally from dawn to midday and sunset, thus requiring regulation mechanismsoperating over a short time-span (minutes, hours) and thus cannot be ac-tivated by the synthesis/degradation of photosynthetic components, whoseeveryday operation would be energetically unsustainable. These short-termprocesses will be described later.

3.3 State 1–State 2 Transitions

State transitions are a protective mechanism, which acts through a reversibleprotein modification in a few minutes timescale [10,11] to balance the energypressure between the two photosystems, physically displacing LHCII antennacomplexes from PSII to PSI [12].

Higher plants and green algae oxygenic photosynthesis works through thesynchronized action of two photosystems: PSII reaction center, whose maxi-mum absorption peak is at 680 nm, and PSI, where it is at 700 nm.

The coordinated and energy collection by the two photosystems is a nec-essary condition for the proceeding of photosynthesis and the generation of astrongly reducing molecule, which is able to transfer its electron to NADP+.

The kinetically limiting step in the electron transport chain between PSIIand PSI involves the oxidation of liposoluble hydrogen carrier plastoquinone(PQH2), which once reduced at the Qb site PSII, diffuses in the thylakoidmembrane to cytochrome b6f complex. In turn, cytochrome b6f is main-tained to an oxidized state by the activity of PSI, which transfers electrons toNADP+. Thus, over-excitation of PSII leads to over reduction of PQ, leavingPSII without a suitable electron acceptor. PSII undergoes charge recombi-nation and closure of energy traps thus increasing Chl lifetime and singletoxygen production.

To avoid this problem, the PQH2 acts as a signal activating chloroplasticprotein kinase [13] to allow phosphorylation of LHCII antenna complex as-sociated to PSII (State I). Phosphorylated LHCII undergoes conformationalchange, disconnection from PSII, and migration to the stroma lamellae whereit associates to PSI (State II). This increases light harvesting, and thus elec-tron transport capacity of PSI thus compensate for the uphill electron carriersover-reduction. Consequence is the oxidation of PQH2 to PQ and downregu-lation of the LHCII kinase. Dephosphorylation of LHCII and its reassociationto PSII reaction centers is ensured by chloroplastic phosphatases. This mecha-nism is photoprotective as it balances a potentially dangerous over-excitationon PSII, source of ROS, and it regulates light-harvesting activity dependingon light intensity and light spectral quality changes consequent to rapid tran-sitions from shade to full sunlight. In fact, the absorption spectrum of PSI isred-shifted with respect to PSII.


In the green alga C. reinhardtii state transitions are the main photopro-tective strategy. This organism is able to shift from PSII to PSI up to 80% ofits LHCII complexes, while in higher plants this parameter is not more than15% [4].

3.4 Carotenoids Play a Fundamental Role in ManyPhotoprotection Mechanisms

The role of carotenoids in photoprotection of photosynthetic systems is ex-tremely important as shown by the early experiments of [15] with carotenoid-less bacterial strains and by the effect of herbicide norfluorazon on plants [16].Carotenoids are 40 carbon atoms polyisoprenoid compounds, made by thecondensation of eight isoprene units. In this class of compounds, we can dis-tinguish between carotenes, linear or cyclic hydrocarbons, and xanthophylls,their oxygenated derivatives. Carotenoids are located in the photosyntheticmembranes, within the chloroplast, both free in the lipid bilayer (1–2%) andbound to the subunits of PSI and PSII in specific binding sites whose oc-cupancy is necessary for the folding of these pigment–protein complexes. Inhigher plants, β-carotene binds to reaction center subunits of both PSI (PsaAand PsaB) and PSII (PsbA,PsbB,PsbC,PsbD). Xanthophylls (lutein, violax-anthin, neoxanthin, and zeaxanthin) are, instead, bound to Lhca and Lhcbproteins forming the outer antenna complexes of PSI and PSII, respectively.In C. reinhardtii, the additional xanthophyll species loroxanthin is present,also bound to Lhc complexes. Location of carotenoid species in both PSI andPSII supramolecular complexes is shown in Fig. 3.2.

Together with β-carotene, xanthophylls act both as photoreceptors, ab-sorbing light energy, which is used in photosynthetic electron transport, andas photoprotectants of the photosynthetic apparatus from excess light energyas well as from the reactive oxygen species that are generated during oxygenicphotosynthesis.

Carotenoids exert their photoprotective role through several mechanisms:(a) Chlorophyll triplet quenching: this action protects thylakoid lipids fromperoxidation by quenching triplet chlorophyll species, which prevents pro-duction of reactive oxygen species (ROS) [17–19]; (b) ROS scavenging: in-complete quenching of 3Chl∗ may yield into ROS production, which can bescavenged by reaction with carotenoids; (c) Activation of heat dissipation ofexcess energy (NPQ): specific xanthophylls (lutein, zeaxanthin) are neededfor operation of nonphotochemical quenching of excess light energy (NPQ),thus decreasing probability for IC and 3Chl∗ formation. All these mechanismsare cooperatively activated for protection of photosynthetic apparatus whenenvironmental conditions promote photooxidative stress.

It is worth nothing that all these mechanisms are rapidly activated:carotenoids detoxify short-living reactive species such as singlet oxygen (1O2),


Core complex: β -caroteneAntenna:

xanthophylls

Antenna: xanthophylls

Core complex: β -caroteneA

B

Fig. 3.2. Location of carotenoid species in both PSI (a) and PSII (b) supramolecularcomplexes. (a) PSI image by [58], based on PSI crystal structure by [59]; (b) PSIIimage by [60]

whose lifetime is ∼200 ns, while the activation of heat dissipation of excess en-ergy (NPQ) requires less than a minute. In fact, carotenoids are involved inmany of the mechanisms that allows survival of the photosynthetic cell whenlight intensity and photosynthetic electron transport rate undergo suddenchanges, thus providing a more efficient response in conditions of repetitiveexcess light exposure, such as under sun flecks deriving from the overcast-ing of the leaf canopy; particularly, those mechanisms that are localized inthe hydrophobic bilayer of the photosynthetic membrane, hosting reactioncenters, and antenna proteins. Additional mechanisms located on the solublephase of the chloroplast stroma are also of great importance for photoprotec-tion and are based on the activity of superoxide dismutase, ascorbate perox-ydase [20]. Their role has been the object of excellent reviews and will not bediscussed here.

The study of xanthophyll function in plants and algae has been carried onin the most recent years through an integrative approach, using genetics, phys-iology, biochemistry, and molecular biology [21]. First of all, it can be askedwhy plants do actually need xanthophylls at all. In fact all the main photopro-tective functions of xanthophylls (namely, triplet quenching and ROS scaveng-ing) are well-known properties of β-carotene [22]. Thus, the extreme conser-vation of carotenoid composition in plant species has little or no explanationon the basis of published chemical properties of these molecules. Nevertheless,the fact that all plant species and to a large extent green algae contain, be-


HO

O

violaxanthin

neoxanthin

lutein

zeaxanthin

O

OH

O

OH

O

HO

HO

HO

OH

OH

Fig. 3.3. The three xanthophylls, namely lutein, violaxanthin, and neoxanthin,contained by all plant species and to a large extent green algae in low light, and thexanthophyll zeaxanthin, synthesized in high light

side β-carotene, the same three xanthophylls, namely lutein, violaxanthin,and neoxanthin (Fig. 3.3) in low light; and that violaxanthin is deepoxidizedto zeaxanthin when light is in excess through the green lineage [23] is a clearbiological demonstration that each of these xanthophylls have a specific func-tion. As mentioned earlier, xanthophylls in the thylakoid membrane are boundto Lhc proteins. Moreover, each Lhc protein has several (2–4) xanthophyll-binding sites. The affinity of each site for xanthophyll species is summarizedin Fig. 3.4 for the major LHCII protein, which constitute by far the mostabundant xanthophyll–protein binding complex in the chloroplast. Detailedbiochemical and spectroscopic analysis has shown that the binding site L1 isoccupied by lutein in all Lhc proteins, site L2 can bind lutein or violaxan-thin, site N1 is specific for neoxanthin, and site V1 bind violaxanthin [24–26].Site L2 can exchange violaxanthin with zeaxanthin, produced in excess light,and undergoes a conformational change, which increases heat dissipation anddecrease the lifetime of the 1Chl∗-excited states [5, 6, 27]. As mentioned ear-lier, xanthophylls have a light harvesting function whose importance can beevaluated by spectroscopic methods. Figure 3.5 shows the deconvolution of


Fig. 3.4. Occupancy of the xanthophyll-binding sites in the major LHCII antennacomplex. The affinity of each site for xanthophyll species is summarized in the figure.From: [45]

1-T spectra and of fluorescence excitation spectra of a typical antenna pro-tein, LHCII. This procedure [28] allows identification of xanthophyll spectralcontributions, while the ratio between the amplitude of each component in1-T vs. fluorescence excitation spectra yields the efficiency of energy transferto Chl a. On these basis, it can be obtained that xanthophylls contribute tolight harvesting by less than 10%, the balance being the effect of Chl a andChl b absorption. This simple consideration suggests that the major functionof xanthophylls is not light harvesting but photoprotection.

3.5 Analysis of Xanthophyll Function In Vivo

Genetic has been used as the main strategy for the understanding of the spe-cific function of each carotenoid species in photosynthesis. Single and multiplemutants targeting gene products catalyzing enzymatic steps in the carotenoidbiosynthesis pathway have been produced by several laboratories yielding col-lection of mutants with altered xanthophyll composition. Most used mutantshave been npq1 (without zeaxanthin), npq2 (without violaxanthin and neox-anthin), lut2 (without lutein), and aba4 (without neoxanthin), which havebeen useful in elucidating the basic function of the different xanthophyllspecies. Ideally, the physiologist would like to work on:


1800000

1600000

1400000

1200000

1000000

A

B

800000

600000

400000

200000

0

1800000

1600000

1400000

1200000

1000000

800000

600000

fluor

esce

nce

exci

tatio

n

400000

200000

0

400 420 440 460

λ (nm)

λ (nm)

1-T

480 500

absa1a2b1b2neoviolalute1lute2fitting

exca1a2b1b2neoviolalute1lute2fitting

520

400 420 440 460 480 500 520

Fig. 3.5. Deconvolution of 1-T spectra (a) and of fluorescence excitation spectra(b) of a typical antenna protein, LHCII. The ratio of the amplitude of each spectralform in (b) vs. (a) yields the efficiency of excitation energy transfer from eachpigment to Chl a, the final sensitizer of photosynthesis. The figure obtained is thatlight absorption by xanthophylls (vs. Chls, 350–750 nm) = 16.8%; contribution toChl a fluorescence by xanthophylls (vs. Chls) in trimeric LHCII = 9.9%

1. Genotypes lacking one specific carotenoid2. Genotypes retaining only one carotenoid species3. Genotypes lacking one or more carotenoid-binding proteins

Although not complete, a considerable collection is now available, whoseanalysis will gave us a detailed functional survey. Mutant collections wereobtained both in higher plants (A. thaliana) and in the green algae modelorganism C. reinhardtii. Table 3.1 includes several genotypes available, withindications relative to laboratories that contributed to their isolation. Thesemutant plants were fundamental for the initial work. Major findings are sum-marized in the following paragraphs.


Table 3.1. State of the art in genetics of carotenoids biosynthesis and function.Mutants affected in carotenoids or in carotenoids-binding proteins composition arelisted

3.6 Nonphotochemical Quenching

In strong light, the energy pressure on the photosynthetic apparatus dramat-ically increases thus causing increase of the fluorescence lifetime of Chl andoxidative stress (see earlier). In higher plants and some green algae, a mecha-nism is activated in these conditions for downregulation of chlorophyll-excitedstates concentration by opening a dissipation channel into heat of the excessenergy absorbed over the rate that can be utilized by photochemistry, whichis saturated in these conditions. Thus, quenching by reaction centers (photo-chemical quenching), when saturared, is integrated by an additional quenchingeffect originated outside reaction centers (NPQ, Nonphotochemical quench-ing) (Fig. 3.1). In doing so, the probability of chlorophyll intersystem crossingis decreased and the formation of reactive oxygen species down rated. NPQcan be measured through the decrease of the leaf fluorescence upon high lightexposure. It was soon realized that NPQ involves several functional compo-nents: state transitions (see 3) account for a small reduction in leaf fluorescence


(called qT); the formation of quenching species within PSII core complex alsocontribute, together with similar effect within antenna system (qI), while thelargest contribution is provided by a quenching mechanism located in the an-tenna proteins, which is activated by the transmembrane pH gradient and is,therefore, called “Energy quenching” or qE.

3.7 Feedback Deexcitation of Singlet-ExcitedChlorophylls: qE

qE is the most relevant NPQ component. It can quench up to 75% of 1Chl∗-excited states with an half time of 1–2 min. qE is triggered by trans-thylakoid∆pH formation: absorption of excess photons causes the buildup of a high∆pH, and the resulting decrease in lumen pH is essential for qE. Thus, qEacts as a feedback-regulated mechanism in photosyntheis, as it is modulatedby the extent of trans-thylakoid ∆pH, generated by photosynthetic electrontransport [29].

The need for stroma/lumen ∆pH and qE is clearly demonstrated by theinhibitory effect on qE of ionophore molecules as nigericin, which is able toabolish pH gradients. Moreover, DCCD (dicyclohexylcarbodiimide), a mole-cule able to covalently bind protonatable residues of proteins, is a powerfulinhibitor of qE [30]. Recently, the target site for DCCD has been localized inthe PsbS subunit [31], a PSII subunit whose deletion abolishes qE [32].

Xanthophylls have a key role in qE. The lut2 mutant, lacking lutein, showsboth a reduced qE amplitude and a slower induction kinetic and an evenstronger effect is observed in the npq1 mutant [3]. In this genotype, the en-zyme violaxanthin deepoxidase (VDE) is knocked out. VDE is usually acti-vated by lumen acidification, thus leading to deepoxidation of violaxanthinto zeaxanthin. This event of light-dependent, reversible deepoxidation of theviolaxanthin pool is referred as “xanthophyll cycle” [33]. Lack of high light-induced Zea synthesis is thus in npq1 mutant and is thus the reason for thestrong reduction in qE. This general figure is supported by the phenotypeof the double mutant npq1lut2, where qE is completely abolished leading tophotooxidative stress [34], and of the npq2 mutant, where zeaxanthin is con-stitutively accumulated, which exhibits qE similar to WT in amplitude butkinetically faster in onset and slower in recovery [3]. These evidences supporta role for zeaxanthin as positive allosteric modulator of qE, expressed uponits reversible binding to Lhc proteins [35].

The mutation npq4 of A. thaliana, the gene encoding the PSII subunitPsbS, is epistatic over carotenoid biosynthesis mutations [32], suggesting thestep it catalyzes is upstream with respect to carotenoid function in qE mecha-nism. Conversely, qE amplitude depends on the stoichiometric amount of PsbSpolypeptide and on the specific protonation of two acidic lumenal residues,E122 and E226, the sites of DCCD inhibition. Inactivation of one of thesesites (A. thaliana point mutants E122Q and E226Q) halves qE, while the


double mutant E122Q-E226Q gives equals the deletion mutant npq4 [31, 36].All together these results suggest that qE mechanism is triggered in excesslight by the protonation of PsbS, acting as a sensor of lumen pH, while thequenching event itself is activated downstream and involves the carotenoidslutein and zeaxanthin. Information on the quenching mechanism itself havebeen obtained by femtosecond transient absorption in vivo [37] suggest thatthe quenching reaction consists into the transient formation of a Chlorophyll–Zeaxanthin radical cation, which then recombines with heat dissipation ofthe excitation energy. This radical cation is proposed to trap excitation en-ergy from PSII chlorophylls (Chl bulk) because of a high rate of formation(0,1–1 ps). Npq4 and npq1 mutants did not show any radical cation forma-tion. A tentative model can be proposed based on the recent finding that (a)PsbS it is not a pigment-binding protein, thus leading to the conclusion thatquenching is not located in this subunit [38]; (b) by the detection of the radicalcations in purified minor chlorophyll-proteins (CP24, CP26, CP29) [61]; and(c) by the finding that zeaxanthin binding induces conformational change andquenching in the very same antenna proteins [27, 39, 40]. Thus, PsbS wouldact as a pH sensor, activated by protonation of E122 and E226 residues,triggering a conformational change in neighbor PSII antenna subunits (CP24,CP26, CP29), whose ability to undergo formation of the radical cation quench-ing species is modulated by their capacity for exchanging violaxanthin withzeaxanthin [41,42].

3.8 ∆pH - Independent Energy Thermal Dissipation (qI)

When zeaxanthin is synthesized in high light and replaces violaxanthin in siteL2 of Lhc antenna proteins, a second type of quenching is produced, which isnot dependent on the formation of radical cations but rather on energy transferfrom Chl a to the short living Zea S1-excited state [43]. This kind of quench-ing, although weaker, is not restricted to the minor proteins CP24, CP26 [27],CP29 [39]; but is also observed in the major LHCII complex [6,43]. A furtherdifference with respect to qE consists in its lack of nigericine sensitivity, oncethat zeaxanthin synthesis has occurred and is thus constitutively active innpq2 mutant, even in the dark [27]. In WT plants, it is induced upon strongillumination that induces accumulation of zeaxanthin. Plants that have previ-ously accumulated zeaxanthin becomes less sensitive to strong light becauseof a molecular shift in their antenna proteins, which are set to their short life-time state through a zeaxanthin-induced protein conformational change thatcan be easily detected through isoelectric point shift and spectroscopy [27,44].

This mechanism, together with qE and qT, contributes to the quenchingphenomenon on the whole called NPQ. Although qE is very fast (t1/2 min=1min) and qT somehow slower (8 min), qI appears to have long relaxation time(20 min–1 h) related to the release of Zea from Lhc proteins. The interplayof these mechanisms ensure photoprotection under changes of light intensitywith different time constants.


3.9 Chlorophyll Triplet Quenching

The above-described mechanisms concur to avoid overexcitation of PSII. Nev-ertheless, by increasing light and/or decreasing temperature or metabolic sinkactivity, overexcitation eventually occurs. As described in section 1, IC (inter-system crossing) converts chlorophyll from its singlet-excited state into triplet,that, ultimately, can react with oxygen yielding ROS. However, carotenoidscan largely prevent reaction with oxygen by quenching 3Chl∗. The transitionenergy level of carotenoid triplet state with nine or more conjugated doublebonds is lower than that of chlorophyll. Carotenoid triplet states, thus, candirectly receive energy from triplet chlorophyll (triplet chlorophyll quench-ing), thus quenching 3Chl∗. This reaction is located within Lhc proteins andprevents the formation of singlet oxygen. Indeed, xanthophylls bound to Lhcproteins are located in proximity to Chl for efficient quenching of 3Chl∗.

As energy of carotenoid triplet state is too low to be transferred to otheracceptor molecules, it is directly dissipated as heat:

3Chl∗ +1 Car →3 Car∗ +1 Chl3Car∗ →1 Car + heat

(3.1)

In LHCII, the major PSII light-harvesting complex, chlorophyll triplet quench-ing is mainly catalyzed by lutein bound in site L1, through transfer of excita-tion energy from nearby chlorophylls [5]. Lutein appears to be the xanthophyllspecies most efficient in 3Chl∗ quenching as determined by direct measure-ments of the kinetics of carotenoid triplet formation upon excitation of Chl.Transient absorption spectroscopy of lutein vs. violaxanthin containing Lhcb1has shown lower Car triplet yield and slower kinetics of Chl a to Car triplettransfer in the latter [45]. Consistently, increased photo-damage has been ob-served in vivo [45] because of formation of 1O∗

2 in chloroplasts and purifiedcomplexes. Thus, changes in the xanthophyll occupancy of sites L1 and L2 ofLhcb proteins, particularly LHCII, affect photoprotection.

In other Lhc proteins, such as Lhca4, also xanthophylls bound to site L2are active in triplet quenching [18,19]. Lhca4 is a subunit of PSI characterizedby red shifted spectra forms absorbing at >700 nm. Although the function ofthese spectroscopic features is the object of debate, analysis of recombinantLhca4 WT vs. mutant missing red-forms showed that xanthophylls efficiencyin triplet chlorophyll quenching was improved in the presence of red-forms.It is proposed that “red” chlorophylls, located near to xanthophyll in siteL2 [46], act as a funnel for Chl triplet states to xanthophyll molecules boundto site L2, thus allowing 100% efficiency in triplet quenching [18].

3.10 Scavenging of Reactive Oxygen Species

Earlier, we have discussed the key role of carotenoids in mechanisms prevent-ing ROS formation. However, these mechanisms can be saturated and ROSformed in vitro and in vivo [47,48]. Carotenoids are well suited as antioxidants


as they are active in scavenging reactive oxygen species generated in thechloroplast. Carotenoids protect from oxidative damage by two general mech-anisms: (1) quenching of singlet oxygen with dissipation of energy as heat;(2) scavenging of radical species thus preventing or terminate radical chainreactions.

Carotenoids are hydrophobic antioxidants located in the thylakoid mem-branes. Xanthophylls bound to the Lhc proteins can catalyze 1O∗

2 scaveng-ing, while β-carotene perform this function in the PS II core complex on1O∗

2produced from interaction of 3P680∗ and O2.After reaction with ROS, the carotenoid is excited to a triplet state

(3Car∗), and then relaxes into its ground state (1Car) by loosing the extraenergy as heat.

1O∗2 +1 Car →3 O2 +3 Car∗

3Car∗ →1 Car + heat(3.2)

Carotenoids can act also as chain-breaking antioxidants in the peroxidationof membrane phospholipids [49] and, therefore, protect unsaturated-rich lipidmembranes from rapid degradation.

It has been reported that A. thaliana npq1 mutant, unable to synthesizezeaxanthin, has higher lipid peroxidation levels than WT; this phenotypeis not related to the lower thermal energy dissipation ability (qE) in npq1because of the absence of zeaxanthin [47].

Differently on mechanisms like qE or qI, where carotenoids active in theprocess are bound to Lhc proteins, an important role in ROS scavengingand detoxification is played by carotenoids, which are free to diffuse in thelipid bilayer. In particular, zeaxanthin has been proposed to favor membranethermostability and protecting from lipid peroxidation [50].

Zeaxanthin has a major role in scavenging: the A. thaliana npq2lut2 doublemutant has zeaxanthin as the only available xanthophyll, and is more resistantthan WT to photooxidative stress and lipid peroxidation [7]. Nevertheless,zeaxanthin-enriched plants have a decreased growth [27] implying that luteinand neoxanthin play a role in the photoprotection against ROS, during normalgrowth in the absence of Zea.

Lutein has a synergistic effect in photoprotection together with zeaxan-thin. A. thaliana lut2 mutant produces more singlet oxygen than WT underphotooxidative conditions; nevertheless, stress symptoms are partially rescuedby enhanced zeaxanthin accumulation. Indeed, plants lacking both zeaxanthinand lutein (npq1 lut2 double mutant) show much more photosensitivity andhigher lipid peroxidation with respect to each single mutant [45]. The sameeffect can be observed in green algae: C. reinhardtii single mutants npq1 andlor1 are altered in photoprotective mechanisms both uphill (singlet and tripletchlorophyll quenching) and downhill (scavenging) ROS production, althoughthey are still able to survive in high light. On the contrary, npq1 lor1 doublemutant shows a lethal phenotype in high light, because of its lower ability to


detoxify singlet oxygen and superoxide anion, as it has been demonstrated bythe exogenous addition of pro-oxidants chemicals [51]. A similar synergisticeffect of zeaxanthin has been shown for neoxanthin. The A. thaliana aba4-1mutant specifically lacks neoxanthin, and is only slightly more sensitive thanWT; however, the double mutants aba4-1 npq1 is prone to photoxidative stresswith respect to npq1, suggesting that Zea compensate for lack of Neo, whosespecific function appears to be the scavenging of superoxide anion (O2) [52]mainly produced in the Mehler reaction.

A role in chloroplast protection from ROS has been recently reportedfor the small amphiphilic lipid tocopherol. Early in vitro experiments havedemonstrated that this compound can efficiently terminate chain reactions ofpolyunsaturated fatty acid free radicals and quench singlet oxygen [53–55].It was found that zeaxanthin-supplemented human cells, in the presenceof either α-tocopherol or ascorbic acid, were significantly more resistant tophotoinduced oxidative stress. The authors postulated that the underlyingmechanism responsible for the synergistic action is based on prevention ofzeaxanthin depletion, because of its free radical degradation. α-tocopherolwould be the final radical scavenger, thus preventing carotenoid consump-tion. This model may be applied also in plants; A. thaliana vte 1 mutantis impaired in tocopherol biosynthesis. When vte1 mutation is coupled withxanthophyll cycle mutation in double mutant npq1 vte1, a higher PSII pho-toinhibition with respect to that of single mutants is reported [56]. These datatogether with the observation that npq1 plants accumulate higher tocopherollevel than WT [47, 57], while vte1 plants accumulate zeaxanthin led to thehypothesis that zeaxanthin and tocopherol have overlapping functions inprotecting from photodamage by ROS.

3.11 Conclusions

Oxygen, essential for animals, is produced by PSII in photosynthetic organ-isms. These organisms use chlorophylls as sensitizers for light-absorptionwater photolysis leading to O2 evolution. The chloroplast is thus a biologicalcompartment where the highest O2 concentrations coexists with reducingspecies produced by light-driven electron transport thus leading to highprobability of ROS formation and photodamage. Mechanisms have evolvedfor preventing formation of photooxidant and detoxify them once they areformed. Photosynthesis heavily depends on their efficiency. First, target ofphotoprotection mechanisms is maintenance of the balance between light ab-sorption and utilization in the ever-changing natural environment; second, thetarget is quenching of Chl triplets when balance is lost. Finally, when ROS areeventually formed, the last target consists in their scavenging. Carotenoids arethe essential molecules for photoprotection in that they are involved in eachof these three levels of photoprotection, namely in thermal energy dissipation(qE and qI), chlorophyll triplet quenching, and direct scavenging of reactiveoxygen species.


References

1. M. Ballottari, L. Dall’Osto, T. Morosinotto, R Bassi, J. Biol. Chem. 282,8947–8958 (2007)

2. B.M. Smith, P.J. Morrissey, J.E. Guenther, JA Nemson, M.A. Harrison,J.F. Allen, A. Melis, Plant Physiol. 93, 1433–1440 (1999)

3. K.K. Niyogi, A.R. Grossman, O. Bjrkman, Plant Cell 10, 1121–1134 (1998)4. T. Morosinotto, R. Baronio, R. Bassi, J. Biol. Chem. 277, 36913–36920 (2002)5. E. Formaggio, G. Cinque, R. Bassi, J. Mol. Biol. 314, 1157–1166 (2001)6. I. Moya, M. Silvestri, O. Vallon, G. Cinque, R Bassi, Biochemistry 40,

12552–12561 (2001)7. M. Havaux, L. Dall’Osto, S. Cuine, G. Giuliano, R Bassi, J. Biol. Chem. 279,

13878–13888 (2004)8. Y. Choquet, F. Zito, K. Wostrikoff, F.A. Wollman, Plant Cell 15, 1443–1454

(2003)9. D.-H. Yang, J. Webster, Z. Adam, M. Lindahl, B Andersson, Plant Physiol 118,

827–834 (1998)10. C. Bonaventura, J. Myers, Biochim. Biophys. Acta. 189, 366–383 (1969)11. J.F. Allen, Curr. Biol. 15, R929–R932 (2005)12. J.F. Allen, Biochim. Biophys. Acta. 1098, 275–335 (1992)13. N. Depege, S. Bellafiore, J.D. Rochaix, Science 299, 1572–1575 (2003)14. R. Delosme, J. Olive, F.A. Wollman, Biochim. Biophys. Acta. 1273, 150–158

(1996)15. W.R. Sistrom, M. Griffiths, R.Y. Stanier, J. Cellular and Comparative Physiol-

ogy 48, 473–515 (1956)16. J.C. Gray, Trends Genet. 19, 526–529 (2003)17. E.J. Peterman, C.C. Gradinaru, F. Calkoen, JC Borst, R. van Grondelle, H. Van

Amerongen, Biochemistry 36, 12208–12215 (1997)18. D. Carbonera, G. Agostini, T. Morosinotto, R Bassi, Biochemistry 44,

8337–8346 (2005)19. R. Croce, M. Mozzo, T. Morosinotto, A. Romeo, R Hienerwadel, R. Bassi, Bio-

chemistry 46, 3846–3855 (2007)20. K. Asada, Annu. Rev. Plant Physiol. Plant Mol. Biol. 50, 601–639 (1999)21. K.K. Niyogi, Annu. Rev. Plant Physiol. Plant Mol. Biol. 50, 333–359 (1999)22. A. Telfer, S. Dhami, S.M. Bishop, D. Phillips, J Barber, Biochemistry 33,

14469–14474 (1994)23. B. ALE Demmig-Adams, K. Winter, A. Kruger, F-C Czygan, in Photosynthe-

sis. Plant Biology Vol.8, ed. by W.R. Briggs, (New York, Alan R. Liss, 1989)pp. 375–391

24. R. Croce, R. Remelli, C. Varotto, J. Breton, R Bassi, FEBS Lett 456, 1–6 (1999)25. R. Croce, S. Weiss, R. Bassi, J. Biol. Chem. 274, 29613–29623 (1999)26. S. Caffarri, R. Croce, J. Breton, R. Bassi, J. Biol. Chem. 276, 35924–35933

(2001)27. L. Dall’Osto, S. Caffarri, R. Bassi, Plant Cell 17, 1217–1232 (2005)28. R. Croce, G. Cinque, A.R. Holzwarth, R. Bassi, Photosynth. Res. 64, 221–231

(2000)29. J.M. Briantais, C. R. Acad. Sci. Hebd. Seances Acad. Sci. D 263, 1899–1902

(1966)


30. A.V. Ruban, R.G. Walters, P. Horton, FEBS Lett. 309, 175–179 (1992)31. X.P. Li, A.M. Gilmore, S. Caffarri, R. Bassi, T Golan, D. Kramer, K.K. Niyogi,

J. Biol. Chem. 279, 22866–22874 (2004)32. X.P. Li, O. Bjorkman, C. Shih, A.R. Grossman, M Rosenquist, S. Jansson,

K.K. Niyogi, Nature 403, 391–395 (2000)33. R.C. Bugos, H.Y. Yamamoto, Proc. Natl. Acad. Sci. USA 93, 6320–6325 (1996)34. K.K. Niyogi, C. Shih, C.W. Soon, B.J. Pogson, D DellaPenna, O. Bjorkman,

Photosynth. Res. 67, 139–145 (2001)35. P. Horton, A.V. Ruban, M. Wentworth, Rev. Philos. Trans. R Soc. Lond. B

Biol. Sci. 355, 1361–1370 (2000)36. X-P Li, A. Phippard, J. Pasari, K.K. Niyogi, Funct. Plant Biol. 29, 1131–1139

(2002)37. N.E. Holt, D. Zigmantas, L. Valkunas, X.P. Li, KK Niyogi, G.R. Fleming, Sci-

ence 307, 433–436 (2005)38. P. Dominici, S. Caffarri, F. Armenante, S Ceoldo, M. Crimi, R. Bassi, J. Biol.

Chem. 277, 22750–22758 (2002)39. M. Crimi, D. Dorra, C.S. Bosinger, E. Giuffra, A.R. Holzwarth, R. Bassi, Eur. J.

Biochem. 268, 260–267 (2001)40. R. Croce, J. Breton, R. Bassi, Biochemistry 35, 11142–11148 (1996)41. R. Bassi, B. Pineau, P. Dainese, J. Marquardt, Eur. J. Biochem. 212, 297–303

(1993)42. T. Morosinotto, S. Caffarri, L. Dall’Osto, R. Bassi, Physiol. Plant. 119, 347–354

(2003)43. T. Polivka, D. Zigmantas, V. Sundstrom, E. Formaggio, G. Cinque, R. Bassi,

Biochemistry 41, 439–450 (2002)44. N.E. Holt, G.R. Fleming, K.K. Niyogi, Biochemistry 43, 8281–8289 (2004)45. L. Dall’Osto, C. Lico, J. Alric, G. Giuliano, M Havaux, R. Bassi, BMC Plant

Biol. 6, 32 (2006)46. T. Morosinotto, J. Breton, R. Bassi, R. Croce, J. Biol. Chem. 278, 49223–49229

(2003)47. M. Havaux, K.K. Niyogi, Proc. Natl. Acad. Sci. USA 96, 8762–8767 (1999)48. A. Krieger-Liszkay, J. Exp. Bot. 56, 337–346 (2005)49. B.P. Lim, A. Nagao, J. Terao, K. Tanaka, T. Suzuki, K. Takama, Biochim.

Biophys. Acta. 1126, 178–184 (1992)50. M. Havaux, Trends Plant Sci. 3, 147–151 (1998)51. I. Baroli, B.L. Gutman, H.K. Ledford, J.W. Shin, B.L. Chin, M. Havaux,

K.K. Niyogi, J. Biol. Chem. 279, 6337–6344 (2004)52. L. Dall’Osto, S. Cazzaniga, H. North, A. Marion-Poll, R. Bassi, Plant Cell 19,

1048–1064 (2007)53. A. Kamal-Eldin, L.A. Appelqvist, Lipids 31, 671–701 (1996)54. P. Di Mascio, T.P. Devasagayam, S. Kaiser, H. Sies, Biochem. Soc. Trans. 18,

1054–1056 (1990)55. M. Wrona, M. Rozanowska, T. Sarna, Free Radic. Biol. Med. 36, 1094–1101

(2004)56. M. Havaux, F. Eymery, S. Porfirova, P. Rey, P. Dormann, Plant Cell. 17,

3451–3469 (2005)57. M. Havaux, J.P. Bonfils, C. Lutz, K.K. Niyogi, Plant Physiol. 124, 273–284

(2000)


58. M.K. Sener, C. Jolley, A. Ben Shem, P. Fromme, N. Nelson, R. Croce,K. Schulten, Biophys. J. 89, 1630–1642 (2005)

59. A. Ben Shem, F. Frolow, N. Nelson, Nature 426, 630–635 (2003)60. A. Eckardt, Plant Cell. 13, 1245–1248 (2001)61. T.J. Avenson, T.K. Ahn, D. Zigmantas, K.K. Niyogi, Z. Li, M. Ballottari,

R. Bassi, G.R. Fleming, J. Biol. Chem. 274, 29613–29623 (2008)

4

Non-Linear Microscopy

D. Mazza, P. Bianchini, V. Caorsi, F. Cella, P.P. Mondal, E. Ronzitti,I. Testa, G. Vicidomini, and A. Diaspro

4.1 Introduction

Non-linear interactions between light and matter have been extensively usedfor spectroscopic analysis of biological, natural and synthetic samples [1–5].In the 1990s, the development in laser technology allowed to apply theseprinciples to the light microscopy field [6–8]. In this context multi-photon exci-tation (MPE) fluorescence microscopy and second harmonic generation (SHG)imaging are representative of the continuing growth of interest in optical mi-croscopy. Although other modern imaging techniques like scanning near-fieldmicroscopy [9], scanning probe microscopy [10] or electron microscopy [11] pro-vide higher spatial resolution, light microscopy techniques have unique char-acteristics for the three-dimensional (3D) investigation of biological structuresin hydrated states, including the direct observation of living samples [12–14].Multi-photon microscopy relies on the property of fluorescent molecules tosimultaneously absorb two or more photons [15]. In this context, the advancesin fluorescence labelling and the development of new fluorescent/luminescentprobes like the so-called quantum dots [16] and the visible fluorescent proteins(VFPs), that can be expressed permanently bound to proteins of interest bygenetically modified cells, allow the study of the complex and delicate re-lationships existing between structure and function in the four-dimensions(x -y-z -t) biological systems domain [17, 18]. MPE shares with the confocalmicroscopy the intrinsical 3D exploration capability and provides some ad-ditional interesting features. First, MPE greatly reduces photo-interactionsand permits to image living samples for long time periods. Second, it allowshigh sensitivity measurements due to the low background signal. Third, sincemost of the fluorescent molecules show a wide two-photon absorption spec-trum, MPE allows simultaneous excitation of multiple fluorescent moleculeswith only one excitation wavelength, reducing the effects of chromatic abber-ations of the optical path. Fourth, two-photon microscopy can penetrate intothick and turbid media up to a depth of some 100 µm. Fifth, MPE can inducechemical rearrangement and photochemical reactions within a sub-femtolitervolume in solutions, culture cells and living tissues.

48 D. Mazza et al.

Another non-linear process that can be applied to the optical microscopyfield is the generation of second harmonic signal. This phenomenon is relatedto the capability of non-centrosymmetrical matter to scatter light at the dou-ble of the illumination wavelength. Since highly organized biological matterssuch as myosin or collagen fiber are excellent SHG sources, SHG allows for3D non-invasive imaging of biological matter in general.

It is important to remember here that other kind of non-linear microscopyhas been developed in the recent years. One example is Coherent anti-StokesRaman (CARS) microscopy; since this signal is directly derived from molec-ular vibrations that are characteristic of the chemical composition and mole-cular structure of the sample, this third-order non-linear technique allows forthe 3D analysis of both the chemical and the morphological information aboutthe sample, without needing of external labelling.

Furthermore, the development of non-linear microscopy favoured pro-gresses of several investigative techniques as fluorescence correlation spec-troscopy [19–22], image correlation spectroscopy [23,24], fluorescence lifetimeimaging [25–28], single molecule detection schemes [29–32], photodynamictherapies [33] and two-photon photoactivation and photoswitching of VFPs[34–36].

Finally, an exciting scenario that has opened in recent years is the non-linear optical nanoscopy [37], related to the possibility of breaking the opticalresolution limit by combining the light coming from different sources, so thatinterference patterns, as in 4Pi microscopy [38], or sequential excitation anddepletion of fluorescent molecules, as in STED microscopy [38], allow to in-vestigate 3D samples at a nanometric level.

4.2 Chronological Notes on MPE

The TPE story starts in 1931, with the theory originally developed by MariaGoppert Mayer in her Ph.D. Thesis [15]. The keystone of TPE theory liesin the prediction that one atom (or molecule) can simultaneously absorb twophotons in the same quantum event, within a temporal window of 10−16–10−15 s. This temporal scale reflects the rarity of a two photon event: in thedaylight an efficient fluorescent molecule undergoes a two-photon absorptiononce every 10 million years, while one photon absorption occurs once a sec-ond [39]. Therefore, to increase the probability of the two-photon event, a highflux of photons is necessary, or in other words a laser source. As for confocalmicroscopy [40], in fact, the development of laser technology has been a keyfactor in the experimental evidence of two-photon events and in the spread-ing of the technique. The first observation of a two-photon signal in CaF2

dates back to 1961 [1], just one year later than the introduction of the firstRuby laser. Some years later, the third-order absorption (three-photon effect)was observed in naphthalene crystals [2]. In the same years second harmonicscattered signal was measured by illuminating quartz crystals with a ruby

4 Non-Linear Microscopy 49

laser beam [41]. For many years the application of these non-linear processeswas limited to the spectroscopic studies of inorganic samples. The first ob-servation of TPE of organic dyes is dated to 1970 [42], while in 1976 Bernsreported a two-photon effect as a result of focusing an intense laser beam ontochromosomes of living cells [43].

The application of these principles to the microscopy field required 20more years. Even if the original idea of generating 3D microscopy imagesby means of non-linear effects was first suggested and attempted in the1970s by Sheppard, Kompfner, Gannaway and Choudhury of the Oxfordgroup [44,45], the practical realization of a ‘two-photon’ microscope is relatedto the pioneering work of W. Denk in W.W. Webb Laboratories (CornellUniversity, Ithaca, NY), which was responsible for spreading the techniquethat revolutionized fluorescence microscopy imaging [6].

4.3 Principles of Confocal and Two-Photon FluorescenceMicroscopy

4.3.1 Fluorescence

The term fluorescence is related to the capability of certain molecules to emitlight (in a time scale of 10−9 s) when they are illuminated with a properwavelength. More precisely, the energy required to prime fluorescence is theenergy that is necessary to produce a molecular transition to an electronicexcited state [46,47]. In other words, if λ is the wavelength of the light deliveredon the sample, the energy provided by photons E = hc/λ (where h = 6.6 ×10−34 J s is the Plank’s constant and c = 3 × 108 m s−1 is the speed of light)should be equal to the molecular energy gap ∆Eg between the ground stateand one vibrational or rotational level of the electronic excited state:

∆Eg = E = hc/λ. (4.1)

Once the molecule has adsorbed the photon, it has several pathways forrelaxing back to the ground state, including non-radiative phenomena, phos-phorescence (associated to the forbidden transition to the triplet state) andfluorescence (Fig. 4.1). In fluorescence the internal conversion from the lowervibrational level of the excited state is not associated with light emission whilethe relaxation to the ground state is achieved by emitting one photon. For thisreason, the fluorescence emission is generally shifted towards a longer wave-length than the one used for exciting the molecule. This phenomenon is knownas Stokes’ shift and it ranges from 50 to 200 nm depending on the fluorescentmolecule in consideration. Conventional imaging techniques use ultraviolet orvisible light for the excitation. In multi-photon excitation the jump betweenthe ground state and the excited one is due to the simultaneous absorption oftwo or more photons [5]. As the sum of the energy of the absorbed photons

50 D. Mazza et al.

Fig. 4.1. Perrin–Jablonski diagram representing the possible pathways for one mole-cule to relax back to the ground state once excited

must match the molecular energy gap, multi-photon processes allow usinglonger excitation wavelength. Multi-photon microscopy generally use infrared(IR) light for the excitation, resulting in the increase of penetration depth inturbid media and in the decrease of the photo-toxicity in biological samples.

Because of the fact that the molecule is memoryless of the way in whichexcitation is accomplished, the two-photon induced fluorescence retains thesame characteristics of conventional emission process. Therefore, in multi-photon excitation, the fluorescent emission is generally at shorter wavelengthsthan the one absorbed by the molecule.

4.3.2 Confocal Principles and Laser Scanning Microscopy

In conventional wide-field microscopy a large portion of the sample is entirelyilluminated with a light source and viewed directly by eye or through anyimage collection system (charged coupled device (CCD), for instance). Withthis method the sample undergoes continuous excitation and all the pointsof the sample, both in-focus and out-of-focus, will contribute to the image.As a consequence, the out-of-focus contribution will appear blurred in theimage, resulting in the decrease of the axial resolution and in the hazing of thecollected image. In this context, as reported by Minsky in 1961, the maximumperformance of a microscopy imaging system should be met if it would bepossible to investigate point by point the observed sample in order to collectat each position the light scattered or emitted by that point alone, rejectingall the contributions from the other parts of the sample, especially from thosebelonging to different focal planes [48]. Even if it is not possible to eliminateevery undesired ray because of multiple scattering, it is straightforward toremove all the rays that do not focus on the point of interest, by using a


point-like source. This can be achieved by using the condenser lens to projecta pinhole aperture (a small aperture in an opaque screen) on the focal planeor by using a laser source focused on the focal plane. In this way the amountof light delivered on the sample is reduced by order of magnitudes withoutaffecting the focal brightness. However, even in this way, all the points alongthe focal axis will be able to scatter light (or to produce fluorescence signal)that will contribute to the image. The solution to this problem resides inplacing a second pinhole aperture in the image plane that lies beyond theback aperture of the objective lens, so that all the light coming from out offocus scattering sources (or fluorescence sources) will be rejected. As shown inFig. 4.2, the final result is a symmetrical set-up, made up of two lenses at thetwo sides of the specimen with two point-like apertures beyond them: in thisway the role played by the lens on the excitation side is identical to the one onthe detection side and their combined effect results in a relevant improvementof the axial response of the imaging system [49]. The strong symmetry ofthe system results also in the possibility (adopted by modern raster scanningconfocal microscopes) of using the same lens both for the excitation and forthe acquisition, in an epi-fluorescence scheme.

Because of the point-like nature of the focused beam, in order to acquiremicroscopic images of an extended portion of the sample, it is necessary toperform a scan. This is usually obtained with one of these methods.

The first one is based on raster scanning (point by point in a line and thenline by line) of the image field. This operation requires a finite time in order

excitation pinhole

light source

specimen

condenser lens(L1)

objective lens(L2)

detection pinhole

detector

foca

l pla

ne

Fig. 4.2. Simplified view of a confocal microscope. A point-like source (a lamp withan excitation pinhole mask or a laser) is projected on the specimen by a condenserlens L1. All the fluorescent molecules illuminated by the cone of light will emit.This emission is brought to the detector by the objective lens L2. Right before thedetector the collection pinhole allow to exclude all the light emitted by out of focusplanes

52 D. Mazza et al.

to collect a sufficient signal from each point of the sample. In the first confocalmicroscopes the scanning was performed by mechanically moving the sampleunder the investigating beam. However, a faster and more effective way ofperforming the raster scanning is related to the use of galvanometric mirrors,which are capable to direct the imaging ray on the different points of thesample. With this method the emitted light is usually detected through aphotomultiplier tube (PMT) and is displayed by memorizing the detectedintensity at each point of the sample. The resulting imaging rate is typicallyof about one image per second, but modern confocal heads with resonantscanning mirrors can scan images up to 10 times faster.

A second possible approach to obtain confocal images consists in usinga disk containing multiple sets of pinholes, namely a Nipkow spinning disk,placed in the image plane of the objective lens [50,51]. A large parallel beamis pointed at the disk and the light passing through the pinholes is focusedby the objective. The light emitted by the sample crosses again the pinholesand it is brought to a high-efficiency and high time-resolution CCD. Whenspinning the disk at a high rate it is possible to produce excitation severaltimes per second, reaching an imaging rate comparable to video rate.

Despite the scanning architecture, both systems allow for optical section-ing: the sample is placed in a conjugate focal plane, and by moving the ob-jective along the optical axis it is possible to scan different fields of view atdifferent depths in the sample and to collect a series of in-focus optical sectionsthat allow for 3D reconstruction.

The quality of the 3D reconstruction will be strictly dependent on thecapability of the system to reject out-of-focus light and, therefore, to thepinhole aperture size: the use of smaller pinholes improves the discriminationof in-focus light and an improvement in axial resolution but will also resultin a lower signal output. However, axial resolution and optical sectioning donot depend only on the pinhole size but also on the optical properties of thelenses (numerical aperture), on the excitation and emission wavelength andon the properties of the sample (its refractive index) as well as on the overalloptical alignment of the instrument.

4.3.3 Point Spread Function of a Confocal Microscope

A simple way for schematizing an optical microscope is a linear space in-variant (LSI) system. In this context it is necessary to determine the re-sponse of the system to an impulsive signal (in this specific case the imageobtained when observing a point-like object). We will refer at this responseas to the point spread function of the microscope (PSF). If the LSI approxi-mation holds, it is possible to idealize any observed object as a collection ofpoint-like sources, and the image can be properly described by the convolu-tion product of the PSF with the object. Let us consider the schematic viewof a confocal microscope that we depicted earlier: a monochromatic point-likesource is focused onto a sample through a lens L1 and the emitted radiation


(also supposed to be monochromatic) is collected through the second lens L2

by a point detector. If hex and hem are, respectively, the impulse response ofthe lens L1 and L2, the radiation distribution delivered on the sample will beUex = (hex ⊗ δs)(x) = hex, where the point-like excitation source has beenmodelled with a Dirac delta.

The fluorescence emitted by each point x of the sample, Uem(x), will beproportional to the product of the field intensity delivered on the sample andon the distribution of the fluorescent dye D(x), Uem(x) = Uex(x)D(x). Thisradiation will be then brought by the second lens to the point-like detector,leading to the signal recorded by the detector, Udet(x) = [(hem ⊗ Uem)δd](x),where the detector has been modelled by a delta function. The overall collectedsignal will be therefore

Itot =∫

Udet(x) dx

=∫

δd(x) dx

∫hem(x − y)Uem(y) dy

=∫

δd(x) dx

∫hem(x − y)hex(y)D(y) dy

=∫

hex(y)D(y) dy

∫δd(x)hem(x − y) dx

=∫

hex(y)hem(−y)D(y) dy. (4.2)

If we consider the particular case of a point-like object, (4.2) provides theimpulsive response of the system, i.e. the total PSF of the confocal microscope.By modelling the point-like object as a Dirac impulse δ0 (4.2) becomes

Itot =∫

hex(y)hem(−y)δ0(y) dy = hex(0)hem(0). (4.3)

If we consider the confocal epi-fluorescence scheme L1 = L2, and if weassume that λex = λem,1 we end up with hex = hem = h. We can generallyextend the previous formulas for an x-y-z scanning coupled to the imagingprocess. We therefore obtain for a general point P (x, y, z), Itot = h2(x, y, z),which is the general expression for the PSF. The mathematical expression forh(x, y, z) can be formulated through the electromagnetic waves scalar theory[52] and through Fraunhofer diffraction, leading to

h(u, v) ∝∣∣∣∣∫ 1

0

J0(vρ) e−iuρ2

2 ρdρ

∣∣∣∣2

, (4.4)

1 The equivalence of the excitation and the emission wavelength is an approxima-tion for fluorescence where generally λex ≤ λem due to Stokes’ shift. A moreprecise expression for the fluorescence case is to consider a weighted mean of theexcitation and emission wavelength λ =

√2λemλex/

√λ2

em + λ2ex.

54 D. Mazza et al.

where J0 is a Bessel function of the 0th order and u and v are dimensionlessvariables that depend on the lens parameters and wavelengths and are respec-tively proportional to the axial and the radial coordinates in the object space:u ∝ z, v ∝ (x2 + y2)1/2.

By making use of the asymptotic expansions of the Bessel function wehave therefore expressions for the points along the optical axis and in thefocal plane:

h(0, v) ∝(

2J1(v)v

)2

h(u, 0) ∝(

sin(u/4)u/4

)2

. (4.5)

The expressions above can be considered lateral and the axial PSF compo-nents of a conventional microscope, as only one lens is used. We can comparethese relation with the confocal PSF:

Itot(0, v) = h(0, v)2 ∝(

2J1(v)v

)4

Itot(u, 0) = h(u, 0)2 ∝(

sin(u/4)u/4

)4

.

(4.6)

In this context the calculation of the full width at half maximum (FWHM)of these expressions represents the system resolution. By limiting our attentionto the axial resolution it can be shown that confocal imaging results in animprovement of a factor 1.4 [53,54]. In particular, the expression of the axialresolution for a pinhole aperture diameter smaller than 1 AU2 results

rz =0.64λ

n −√

n2 − NA2(4.7)

where n is the refractive index of the medium and NA is the numericalaperture of the objective.

Despite the theoretical formalism for evaluating the ideal PSF of a confocalsystem shows the improvement in lateral and axial resolution, some considera-tion must be given about the real response of a system when imaging non-idealsamples. First it must be considered that the pinhole aperture is strictly re-lated to the improvement in resolution: by opening the pinhole, the detectorcannot be considered anymore as punctual and the resolution gets worse.

Therefore, in the observation of dim samples some compromise betweenthe resolution and the intensity of the collected signal must be found. Table 4.1reports the dependence of axial and lateral resolution on the pinhole size. Theexperimental results have been obtained by imaging sub-resolved fluorescentbeads (Polyscience, diameter (64 ± 9 µm)) with a × 100, 1.3 NA objectiveunder 488 nm excitation.

Furthermore, the real PSF will depend on the physical and optical charac-teristics of the observed sample. In particular, the refractive index mismatch2 AU is the Airy-Unit, the diameter of the Airy disk.


Table 4.1. FWHM of confocal PSFs for different pinhole sizes

Oil (n = 1.5)

Lateral (nm) Axial (nm)

Pinhole 20 µm Pinhole 50 µm Pinhole 20 µm Pinhole 50 µm

Experimental 186 ± 6 215 ± 5 489 ± 6 596 ± 4Theoretical 180 210 480 560

Table 4.2. FWHM of confocal PSFs for different focusing depths and objectivemedia

Air Glycerol OilDepth Lateral Axial Lateral Axial Lateral Axial(µm) (nm) (nm) (nm) (nm) (nm) (nm)

0 187 ± 8 484 ± 24 183 ± 14 495 ± 29 186 ± 6 489 ± 630 244 ± 10 623 ± 9 221 ± 5 545 ± 12 197 ± 10 497 ± 2160 269 ± 11 798 ± 10 252 ± 7 628 ± 9 186 ± 12 196 ± 1990 277 ± 5 1063 ± 24 268 ± 8 797 ± 26 191 ± 9 484 ± 12

between the objective immersion medium and the sample solution can playa crucial role as some spherical aberration effects can arise, leading to a lossof axial resolution and deforming the actual shape of the PSF [55]. In par-ticular, the effects of spherical abberation get more important when focusingdeep in the sample. Table 4.2 reports the broadening of the lateral and axialresolution when imaging planes at different depths of focus.

4.4 Two-Photon Excitation

Let us consider the special case of two-photon excitation. All the consid-erations made here can be easily extended to the more general case ofmulti-photon excitation. In TPE, two photons are absorbed by a fluorescentmolecule in the same quantum event. The interaction may occur only if theglobal energy delivered by the photons is identical to the energy gap of themolecule between the ground state and some vibrational level of the excitedelectronic state, or in other words if the sum of the energy of the two photons(that do not have to be necessarily identical) is equal to the energy requiredfor prime excitation with a conventional one-photon absorption (ref. Fig. 4.3).This means in terms of the wavelength of the two photons that prime thenon-linear excitation, λ1 and λ2:

λ1P =(

1λ1

+1λ2

)−1

, (4.8)

56 D. Mazza et al.

Fig. 4.3. Perrin–Jablonsky simplified diagram representing the differences betweenconfocal and two-photon excitation microscopy. Fluorescence emission does not de-pend on the way in which excitation is performed

where λ1P is the wavelength required to induce one-photon excitation. Forpractical purposes the wavelength of the two photons is chosen identical, sothat

λ2P = λ1 = λ2 = 2λ1P and ∆Eg =2hc

λ2P. (4.9)

Considering the two-photon process as non-resonant, we can assume theexistence of a virtual intermediate state; the electron will reside in the virtualstate for a time, τvirt, that can be calculated using the time–energy uncertaintyprinciple:

∆Egτvirt � h/2 → τvirt � 10−15 − 10−16 s. (4.10)

The two photons will have to be delayed no more than τvirt to inducethe non-linear interaction. It is clear that for a high flux of photons it istherefore required to have the two-photon interaction. In TPE the photonsspatial and temporal concentration flux will have a crucial role. The shorterwavelength required for TPE allows using near IR to excite UV and visibleelectronic transitions. The rigorous development of multi-photon theory re-quires to apply perturbation quantum theory [56]. In particular, by solvingthe perturbation expansion of time-dependent Shrodinger equation and by us-ing a Hamiltonian containing the dipole interaction of the molecule with theincoming radiation, it is possible to show that the first-order solution corre-sponds to one-photon interaction, while higher order solution are the n-photonones [57]. Such a derivation allows to show that TPE depends on the squareof the intensity delivered on the molecule; however, the same result can beobtained with classical or semi-quantic considerations [58].


Therefore, the fluorescence intensity emitted by one molecule can be con-sidered proportional to the two-photon molecular cross section δ2(λ) and tothe square of the intensity delivered on the sample:

If(t) ∝ δ2I(t)2 ∝ δ2P (t)2[π

(NA)2

hcλ

]2

, (4.11)

where P (t) is the laser power, and the intensity of the incoming radiation iscalculated by using the paraxial approximation in an ideal optical system.Two-photon excitation is generally induced with ultra-fast pulsed lasers. Thetime averaged intensity emitted by one molecule over a time T will be therefore

〈If(t)〉 =1T

∫ T

0

If(t) dt ∝ δ2

[π

(NA)2

hcλ

]2 1T

∫ T

0

P (t)2 dt. (4.12)

We can now consider fp = 1/T as the repetition rate of the pulsed laserand τp as the pulse width. If we consider that the emission of the laser isdescribed by the profile

P (t) = Pmax for 0 < t < τp

P (t) = 0 for τp < t < T, (4.13)

then the mean square power delivered on the sample P 2ave = 1/T

∫ T

0P (t)2 dt

becomes P 2ave = P 2

maxτpfp and (4.12) becomes

〈If(t)〉 ∝ δ2P 2

ave

τpfp

[π

(NA)2

hcλ

]2

. (4.14)

Writing the same relationship for a continuous wave (CW) laser, we obtain

〈If(t)〉 ∝ δ2P2ave

[π

(NA)2

hcλ

]2

. (4.15)

Comparing (4.13) and (4.15) it is evident that the excitation of moleculeswith a pulsed laser is more efficient: in order to obtain the same fluorescenceemission with a CW laser, it is necessary to use an average power 1/

√τpfp

higher relatively to the pulsed laser case.From (4.13) it is possible to easily evaluate the probability na for a fluo-

rophore to absorb two photons simultaneously during a single pulse:

na ∝ δ2P 2

ave

τpf2p

[π

(NA)2

hcλ

]2

. (4.16)

The selection of the repetition rate and of the width of the beam pulseis related to the necessity of avoiding saturation of the fluorescence when na

approaches unity. In other words, during the beam pulse the molecule must

58 D. Mazza et al.

have enough time to relax back to the ground state as this is a prerequi-site for the absorption of another pair of photons. If not, saturation effectsarise, leading to a worsening of the axial and radial resolution of the sys-tem [59]. The application of (4.16) allows to choose for proper optical andlaser parameters that maximize excitation out of the saturation regime. Inparticular, the typical values for the ultra-fast lasers used in two-photon mi-croscopy have τp = 80−150 fs and fp = 80−100MHz. The last considerationsmust be given about the two-photon cross section δ2. Because of the fact thatthe quantum-mechanical selection rules for two-photon excitation differ fromtheir one-photon counterpart, it is not easy to extend the data for conven-tional absorption to the non-linear case even if a simple “rule of thumb” can beconsidered. In fact, in general we can expect to have a two-photon excitationpeak at a wavelength that is the double of the one-photon excitation maxi-mum. However, the knowledge of the global trend of two-photon cross-sectionfor a particular molecule requires the direct measurement. Figure 4.4 showsthe TPE cross section for some of the most common fluorescent molecules.Because of the wide nature of the excitation spectra it can be noted thatone single wavelength can be used for the simultaneous excitation of multipledyes [60, 61]. It has been showed that endogenous fluorescent molecules likeflavoproteins, NAD(P)H and tryptophane exhibit TPE fluorescence [61, 62]and also the fluorescent proteins like GFP (green fluorescent protein) arecapable of undergoing TPE [63–65]. The two-photon cross-sections are gen-erally expressed in GM (Goppert-Mayer, 1GM = 10−58m4 s). GFP variants

600

TI: SapphireSHG of Cr: YAG

SHG of Cr: Forsterite

Cr: LiSGAFCr: LiSAF

Nd: YLF or Nd:glass

10−3

10−2

10−1

100

102

103

10

700 800

Pyrene Cascade Blue

Lucifer Yellow

Fluorencein

Bodipy FI

Coumarin

Rhodamine B

Dil

DAPI

Bis-MSB

Dansyl

900

Wavelength (nm)

Tw

o ph

oton

Exc

itatio

nC

ross

Sec

tions

(G

M)

1000 1100

Fig. 4.4. Exemplary two-photon cross-sections for some typical fluorophores forbiological imaging. The bars represent typical emissions of commonly used lasersources for TPE


have cross sections between 4 and 60 GM [66]. As comparison we can con-sider NADH that at its excitation maximum, 700 nm, shows a cross sectionof 0.02 GM [61]. Quantum dots can show cross sections up to 2,000 GM.

4.5 Two-Photon Optical Sectioning

In the following we will describe how two-photon (or multi-photon) excitationallows to spatially control excitation along the optical axis, limiting the ex-citation volume to a sub-femtoliter volume, thus allowing optical sectioning.We discussed above that the fluorescence signal emitted under two-photonexcitation will depend on the second power of the excitation intensity. We cantherefore consider that the intensity delivered point by point on the sample,Iex, is proportional to the lens PSF, which in the paraxial approximation isdescribed by (4.4). Therefore, we can expect that the fluorescent signal emit-ted by the sample Iem(x, y, z) will be proportional to I2

ex and for the axial andradial components of Iem – introducing the dimensionless variables u, v – weget [67]

Iem(0, v)=h(0, v)2 ∝(

2J1(v)v

)4

Iem(u, 0)=h(u, 0)2 ∝(

sin(u/4)u/4

)4

.

(4.17)

These expressions are identical to the axial and radial parts of the totalPSF of a confocal system (ref. (4.6)). However, it must be noted that whileconfocal microscopy achieves this result at the detection side, by means of thepinhole, in two-photon excitation the dependence of the fluorescence intensityon the inverse of the 4th power of the axial coordinate is achieved directlyin the excitation step. This means that while in confocal microscopy a thickvolume of the sample is excited and the selection of the in-focus contribution isobtained through the pinhole at the detection level, in two-photon excitation,only the molecules in a small volume (order of the femtoliter) around the focalpoint are properly excited and emit fluorescence [68,69]. This fact also impliesthat the contributes far-off the focal plane (depending on the NA of the lens)will not be affected by photobleaching [70,71] or phototoxicity [72–74] and donot contribute to the signal detected if a TPE architecture is used. In thisway we have no need for a pinhole as all the fluorescent emission is supposedto be originated at the focal plane.

This means that TPE is intrinsically three-dimensional as it allows foroptical sectioning by collecting images of the sample plane-by-plane and thenreconstructing the three-dimensional map of the emitted fluorescence (ref.Fig. 4.5). It must be underlined that since all the fluorescence collected is nec-essarily originated at the focal point, the efficiency of the collection of thesignal is much higher than in the confocal case. In TPE, in fact, over 80% ofthe signal arise from a 700–1,000 nm (depending on the NA of the objective)thick region around the focal plane [54, 75]. This results in a reduction in

60 D. Mazza et al.

Fig. 4.5. Exemplary optical sectioning obtained with two-photon excitation at720 nm. The sample is Colpoda cists, and the signal is produced by autofluores-cence of the membranes

background and an increase in the image contrast. This compensates the de-crease in resolution due to the longer wavelength compared to the one usedin conventional excitation (4.7). However, the use of infrared wavelength in-stead of UV–visible allows deeper penetration as the long wavelength usedin TPE (and in general in MPE) will be scattered less than the UV–visiblelight [55, 76]. Because of the high intensity delivered in the focal point (com-pared to conventional excitation), it must be noted that the laser power mustbe finely controlled in order to prevent photo-damage effects [73]. On theother side, the confinement of the excitation processes allows to photochem-ically modify the sample properties with a high 3D spatial control, openingnew frontiers in the context on micro- and nano-surgery [77, 78]. Finally, thelocalization of the photobleaching effect allows to acquire 3D maps of thesample for longer times than in conventional and in confocal architecture, aswhen observing a specific slice of the sample, the out-of-focus contributes willnot be affected by photobleaching [70].

4.6 Two-Photon Optical Setup

The basic setup for multi-photon imaging includes the following elements:a high-intensity ultra-fast infrared laser source (femto- or picosecond pulsewidth with a 80–100 MHz repetition rate); a laser beam scanning system


(i.e. a pair of galvanometric mirrors); high numerical aperture objectives (NA1.0–1.4) in order to deliver high peak intensity in the focal point; and a highlyefficient detection system.

Laser source plays an essential role in the production of MPE signal. Be-cause of relatively low cross-section of the non-linear processes, high photonflux is required, >1024 photons cm2 s−1 at the focal plane [79]: considering thespectral range of 600–1,100 nm this means a peak intensity in the MW cm−2–GW cm−2. By using an ultra-fast pulsed laser in combination with a highnumerical aperture objective, this results in a mean laser power of 50 mW orless. This allows for a sufficient peak intensity to induce TPE, with a meanpower levels that are biologically tolerable [72,80].

The most common ultra-fast pulsed lasers are Ti:sapphire femtosecondlaser sources [40], which can be tuned in wavelength between 700 and 1,050 nmallowing most of the common fluorescent molecules to be excited in TPEregime. Typical parameters for Ti:sapphire lasers include an average power of700 mW–1 W, pulse width of 100–150 fs and repetition rate of 80–100 MHz.Other laser sources for MPE include Cr-LiSAF, pulse-compressed Nd-YLF inthe femtosecond regime, and mode-locked Nd-YAG. Picosecond pulse widthcan also be reached by using a pulse stretcher. It must be noted that due totime–energy uncertainty principle, ∆E∆T ≥ h/2, a shorter pulse width willreturn in a broader emission in terms of wavelengths: by considering ∆T = τp,∆λ/λ = ∆E/E and E = hc/λ, we obtain

∆λ =λ2

2πτpc, (4.18)

which for a central laser emission wavelength of 1,000 nm and pulse width τp =100 fs gives an uncertainty on the wavelength of about 5 nm. Furthermore, itmust be considered that, when the laser light crosses the sample, a temporalbroadening of the pulse occurs. Direct measurement of the pulse width atthe focal plane is not an easy task [81–83]. For practical purposes, it can beassumed that at the focal volume 1.5–2 times broadening is obtained.

Some notes also must be given to the scanning system. Generally, a x-yraster scan is performed by means of galvanometric mirrors [84]. Therefore,the image acquisition rate is generally limited by the mechanical propertiesof the scanner. Fast scanning methods based on resonant scanners can beperformed as well; however, it must be considered that fast scanning and sub-sequent high temporal resolution may require some compromises about spatialresolution and sensitivity, since by scanning faster the laser will illuminate forshorter times each point of the sample, resulting in a lower collected signal.Axial scanning can be performed by a variety of solutions, the most commonbeing a single objective piezo nanopositioner or a galvanometric object table.Two-photon architecture usually allows to switch between confocal and TPEimaging retaining the x-y-z focal position [85]. Furthermore, the use of electro-optical (like electro-optic modulators, EOMs) or acusto-optical devices allow

62 D. Mazza et al.

Excitationlight

Scanning system

Non-descanned detector

DM

Flu

ore

scen

ce

Detector

Detector

DM

X-YScanner

Pinhole

Objectivelens

FocalPlane

Scattering specimen

Fig. 4.6. Laser sources in use at LAMBS Microscobio research center of theUniversity of Genoa for TPE

to rapidly modulate the laser intensity: in this way it is possible to defineregions to be imaged with different powers during the very same scan.

As depicted in Fig. 4.6 it is possible to consider two popular approachesto TPE microscopy, the descanned and non-descanned modes. The first con-figuration uses the same optical pathway as used in confocal microscopy, andthe light emitted by the sample is deflected back by the scanning mirrors andbrought to the photomultipliers. The pinhole is usually kept open – or removed– as it is not necessary for TPE optical sectioning. In the non-descanned modethe microscope architecture is optimized for collecting the maximum TPEsignal: pinholes are removed, the emitted light is collected through a dichroicmirror and it does not cross the scanning system. [61,76,86]

Despite the configuration used, the fluorescent light is brought by thedetection system by a dichroic mirror. Several types of detectors can be used,


Fig. 4.7. Optical scheme of two-photon microscope in descanned and non-descannedconfiguration (Courtesy of M.Cannel and C. Soeller)

including PMTs, CCD cameras and avalanche photodiodes. PMTs are themost commonly used for their low cost, good sensitivity in the visible range(the quantum efficiency is about 20%–40% in the blue green range, droppingdown to 1% in the red) and for the large photosensitive area that allows a gooddynamic range; thus an efficient collection of the signal (Fig. 4.7). Avalanchephotodiodes are extremely efficient in terms of sensitivity (quantum efficiencyof about 70%–80%). Unfortunately, they are pretty expensive and their smallphotosensitive area requires the descanned architecture, limiting the globalsignal collection efficiency. CCD cameras are used for fast imaging, and areparticulary useful in the context of tandem or multi-focal imaging. Finally,regarding the collection of the signal, it must be noted that due to the factthat the fluorescent signal is generated only in the focal point, we could thinkof an ideal TPE microscope with a detector that is able to collect all the lightcoming from the sample, as the fluorescence signal can be easily recognizedby the scattered excitation light by means of the wavelength – i.e. with alow-pass filter.

4.7 Second Harmonic Generation (SHG) Imaging

In conventional microscopy, contrast methods consist of phenomena like ab-sorption, reflection, scattering and fluorescence: in this conditions the speci-men response is linearly dependent with the incoming light. This means thata linear relationship exists between the electric field strength E of the lightand the induced polarization of the object P . In particular, if we consider anincoming oscillating field with a non-resonant frequency (i.e. a frequency thatis not absorbed by the molecules of the material), the optical response can beapproximated to be a first-order response:

P (t) = ε0χ(1)E(t), (4.19)

64 D. Mazza et al.

where ε0 is the dielectric constant of the vacuum and χ(1) is the linearsusceptibility of the specimen. Moving to the non-linear domain, high powerintensity cause a large variety of unusual responses, with a non-linear depen-dency on the applied electric field. Second harmonic generation is one of thesenonlinear optical effects in which the incident light is coherently scattered bythe specimen at twice the optical frequency and at certain angles [87]. We cangenerally write the non-linear correspondent of (4.19) as

P (t) = ε0

(χ(1)E(t) + χ(2)E2 + χ(3)E3...

), (4.20)

where the first left term on the right is the linear scattering, the second isrelated to SHG, the third to third harmonics, etc. In the above equation,because the fields are vectors, the nonlinear susceptibilities are tensors. Aseach atom acts as an oscillating dipole that radiates in a dipole radiationpattern, the radiation phase among the enormous number of atoms must bematched to induce constructive interference and thus non-linear generation isallowed under phase-matching conditions (i.e. when the scattered light is inphase).

For the same reason, only molecules that exhibit a specific symmetry.In particular, only non-centrosimmetric materials can originate SHG. TheSHG signal depends strictly on the relative orientation between the polar-ization of the incoming light and the direction of the symmetry constraints;a polarization analysis of the second harmonic signal can provide useful in-formation about the orientation of molecules, impurities in crystal structuresand characteristics of surfaces and optical interfaces. As for linear scattering,second harmonic generation is not associated with absorption and involvesonly virtual state transitions that are related to the imaginary part of thenonlinear susceptibilities, and so no energy is deposited in the specimen andno damage can be produced. This is one of the major advantages of applyingSHG to the microscopy field and in particular to the investigation of biologicalsamples [88, 89]. Furthermore, SHG imaging preserve the intrinsical capabil-ity of 3D investigation of matter, since the high photon flux required forgenerating this non-linear signal is achieved only within a femtoliter volumearound the focal point of the lens. Since both SHG and TPE can be observedsimultaneously from the same sample, the correlative analysis of these twosignals provide additional insight about the specimen, allowing not only toidentify the molecular source of the SHG, but also to probe radial and lateralsymmetry within structures of interest.

Recognition of the SHG relies on the property that the emitted light hasdouble wavelength of the incoming radiation. Therefore, by changing the colorof the illumination laser we expect to observe an analogue shift in the emis-sion wavelength. The emission of SHG signal is not isotropic, as it will bemore efficient in the forward direction. However, in dense samples, multiplescattering allow for a detection in the backward direction too. Therefore, SHGcan be usually collected both along the transmitted light and along the epi-pathways of the microscope.


A number of plant structures are capable of efficiently generating secondharmonic signals. These structures include stacked membrane structures suchas grana, starch granules, secondary cell wall, cellulose, cuticle and cuticularwaxes, and silica deposits (bio-opals). Starch granules exhibit high conver-sion efficiency in SHG; in fact, a piece of potato tuber placed in an unfocused,ultra-fast laser beam can efficiently generate a bright SHG beam in the forwarddirection. For example, the SHG signal from a potato (Solanum tuberosumL.) starch granule is so strong that it is visible to the naked eye even underambient room light. Cellulose is a linear molecule without branching. Neigh-boring cellulose chains may form hydrogen bonds leading to the formation ofmicro-fibrils (20–30 nm) with partially crystalline parts called micelles. All thehighly organized structure may be responsible for strong SHG signal such ascollagen or myosin fibers. SHG allows therefore for three-dimensional imagingand microscopical investigation of tissue organization with typical confocalresolution, without the need for any external labeller [90–92]. The possibilityof selecting the wavelength in the IR range permits to perform practicallynon-invasive in vivo deep tissue imaging, being a promising tool for skin en-doscopy.

4.8 Conclusions

The investigation of soft and living matter requires instruments capable toinvestigate the relationships between the 3D structure of the specimen andits functions and activity. Optical microscopy and especially the advancesbrought by confocal and non-linear microscopy in the last decades offers apowerful tool for this aim. Confocal microscopy allows to generate 4D (x-y-z-t)views of the specimen in a non-invasive way, preserving the vital condition ofthe sample, with a strong reduction of the out-of-focus haze, allows the multi-ple observation of different dyes by spectral fingerprinting and recent scanningheads provide a good tool for visualizing at video rate fast in vivo dynam-ics. The advances in confocal microscopy, together with the development ofnew fluorescent labellers and the spreading of visible fluorescent proteins havefavoured the development and the spreading of two-photon and multi-photonimaging. Compared to confocal microscopy, we can mainly recognize two im-portant properties by MPE.

1. The MPE excitation is confined both in the x-y directions and alongthe optical axis. Therefore, no fluorescence signal arises from elementsout of the focal point. Consequently, optical sectioning can be performedwithout the use of a pinhole or deconvolution algorithms. Furthermore,photo-bleaching and photo-damage are also confined to the focal point,and the signal-to-noise ratio increases as well as the image contrast. Thedevelopment of proprietary schemes, like the non-descanning acquisition,further improves the efficiency of the signal collection.

66 D. Mazza et al.

2. The use of near-infrared wavelengths allows to minimize biological damageas biological tissues in general poorly absorb infrared light. This combinedwith the in-focus photobleaching allows for long-term observation of liv-ing samples. Because of the wide two-photon cross section of most of thefluorescent molecules, it is possible to excite different dyes at once withonly one laser line, reducing the average power delivered on the sample.Furthermore, the use of IR light permits deeper penetration into the sam-ple (up to 0.5 mm) as both absorption and scattering are reduced withthis wavelength. Finally, because of the fact that the separation betweenexcitation and emission (excitation in the IR, emission in the UV–visible)is usually wider than in conventional excitation, it is considerably eas-ier do discriminate between the actual emission and scattered/reflectedexcitation light.

The advances of TPE is strictly connected to the developments in an-other non-linear microscopy technique, second harmonic generation imaging.It offers a practically non-invasive tool for deep tissue imaging as it does notrequire fluorescent labelling of the samples and it retain the 3D investigationand optical sectioning properties of multi-photon excitation. Also in this caseall the advantages using IR light for the illumination are conserved.

Since relevant generation of second harmonic signal is associated with thepresence of tissue constituents as collagen, SHG may play an important rolein skin disease diagnosis and in the medical research field in general, being apromising endoscopy tool.

The combination of MPE microscopy and SHG imaging provides botha deep insight into biological matter and offers possibilities in the nano-manipulation of living cells and tissues and in the 3D micro- and nano-surgeryfield. The range of application of these techniques is rapidly increasing in thebiomedical, biotechnological and biophysics sciences and it is now facing theclinical applications.

References

1. W. Kaiser, C.G.B. Garrett, Phys. Rev. Lett. 7, 229 (1961)2. S. Singh, L.T. Bradley, Phys. Rev. Lett. 12, 162 (1964)3. D.M. Friedrich, J. Chem. Educ. 59, 472 (1982)4. D.M. Friedrich, W.M. McClain, Annu. Rev. Phys. Chem. 31, 559 (1980)5. P. R. Callis, Annu. Rev. Phys. Chem. 48, 271 (1997)6. W. Denk, J.H. Strickler, W.W. Webb, Science 248(4951), 73 (1990)7. A. Diaspro, IEEE Eng. Med. Biol. Mag. 15, 29 (1996)8. A.J. Koster, J. Klumperman, Nat. Rev. Mol. Cell. Biol. (Suppl), SS6 (2003)9. U. Durig, J.K. Gimzewski, D.W. Pohl, Phys. Rev. Lett. 57(19), 2403 (1986)

10. G. Binnig, C.F. Quate, Ch. Gerber, Phys. Rev. Lett. 56(9), 930l (1986)11. E. Ruska, M. Knoll, Z. Tech. Phys. 12, 389 (1931)12. J.B. Pawley (ed.), Handbook of Biological Confocal Microscopy. (Plenum, New

York, 2005)


13. A. Periasamy (ed.), Methods in Cellular Imaging. (Oxford University Press,New York, 2001)

14. A. Diaspro (ed.), Confocal and Two-Photon Microscopy: Foundations, Appli-cations and Advances. (Wiley, New York, 2002)

15. M. Goppert-Mayer, Ann. Phys. 9, 273 (1931)16. J.K. Jaiswal, E.R. Goldman, H. Mattoussi, S.M. Simon, Nat. Methods 1(1),

73 (2004)17. D.J. Arndt-Jovin, M. Robert-Nicoud, S.J. Kaufman, T.M. Jovin, Science

230(4723), 247 (1985)18. J. Lippincott-Schwartz, E. Snapp, A. Kenworthy, Nat. Rev. Mol. Cell. Biol.

2(6), 444 (2001)19. K.M. Berland, P.T. So, E. Gratton, Biophys. J. 68(2), 694 (1995)20. P. Schwille, U. Haupts, S. Maiti, W.W. Webb, Biophys. J. 77, 2251 (1999)21. K.G. Heinze, M. Jahnz, P. Schwille, Biophys. J. 86, 506 (2004)22. Q. Ruan, M.A. Cheng, M. Levi, E. Gratton, W.W. Mantulin, Biophys. J. 87,

1260 (2004)23. P.W. Wiseman, J.A. Squier, M.H. Ellisman, K.R. Wilson, J. Microsc. 200, 14

(2000)24. P.W. Wiseman, F. Capani, J.A. Squier, M.E. Martone, J. Microsc. 205, 177

(2002)25. T. French, P.T. So, D.J. Weaver, T. Coelho-Sampaio, E. Gratton, E.W. Voss,

J. Carrero, J. Microsc. 185(Pt 3), 339 (1997)26. K. Konig, S. Boehme, N. Leclerc, R. Ahuja, Cell. Mol. Biol. 44, 763 (1998)27. J. Sytsma, J.M. Vroom, C.J. De Grauw, H.C. Gerritsen, J. Microsc. 191, 39

(1998)28. M. Straub, S.W. Hell, App. Phys. Lett. 73, 1769 (1998)29. J. Mertz, C. Xu, W.W. Webb, Opt. Lett. 20, 2532 (1995)30. X.S. Xie, H.P. Lu, J. Biol. Chem. 274, 15967 (1999)31. G. Chirico, F. Cannone, S. Beretta, G. Baldini, A. Diaspro, Microsc. Res. Tech.

55(5), 359 (2001)32. G. Chirico, F. Cannone, G. Baldini, A. Diaspro, Biophys. J. 84(1), 588 (2003)33. J.D. Bhawalkar, N.D. Kumar, C.F. Zhao, P.N. Prasad, J. Clin. Laser Med.

Surg. 15(5), 201 (1997)34. G.H. Patterson, J. Lippincott-Schwartz, Science 297(5588), 1873 (2002)35. G. Chirico, A. Diaspro, F. Cannone, M. Collini, S. Bologna, V. Pellegrini,

F. Beltram, Chem. Phys. Chem. 6(2), 328 (2005)36. M. Schneider, S. Barozzi, I. Testa, M. Faretta, A. Diaspro. Biophys. J. 89(2),

1346 (2005)37. S.W. Hell, Nat. Biotechnol. 21, 1347 (2003)38. S.W. Hell, M. Schrader, H.T. van der Voort, J. Microsc. 187(Pt 1), 1 (1997)39. W. Denk, K. Svoboda, Neuron 18(3), 351 (1997)40. E. Gratton, M.J. van de Ven, in Handbook of Confocal Microscopy, ed. by

J.B. Pawley (Springer, New York, 1995), p. s6941. P.A. Franken, A.E. Hill, C.W. Peters, G. Weinreich, Phys. Rev. Lett. 7, 118

(1961)42. P.M. Rentzepis, C.J. Mitschele, A.C. Saxman, Appl. Phys. Lett. 17, 122 (1970)43. M.W. Berns, Biophys. J. 16(8), 973 (1976)44. J.N. Gannaway, C.J.R. Sheppard, Opt. Quant. Electron. 10, 435 (1982)45. C.J.R Sheppard, A. Choudhury, Opt. Acta 24, 1051 (1977)

68 D. Mazza et al.

46. J.B. Birks, Photophysics of Aromatic Molecules (Wiley, London, 1970)47. C.R. Cantor, P.R. Schimmel, Biophysical Chemistry. Part II: Techniques for

the Study of Biological Structure and Function (Freeman, New York, 1980)48. M. Minsky, Scanning 10, 128 (1988)49. C.R.J. Sheppard, in Confocal and Two-Photon Microscopy: Foundations, Ap-

plications and Advances, ed. by A. Diaspro (Wiley, New York, 2002), p. 150. M. Petran, M. Hadravsky, M.D. Egger, R. Galambos, J. Opt. Soc. Am. 58,

661 (1968)51. G.S. Kino, T.R. Corle, Phys. Today 42, 55 (1989)52. M. Born, E. Wolf, Principles of Optics (Pergamon, New York, 1993)53. G.J. Brakenhoff, E.A. van Spronsen, H.T. van der Voort, N. Nanninga, Meth-

ods Cell. Biol. 30, 379 (1989)54. T. Wilson, C.R.J. Sheppard, Theory and Pratice of Scanning Optical Mi-

croscopy (Academic Press, London, 1984)55. A. Diaspro, F. Federici, M. Robello, Appl. Opt. 41(4), 685 (2002)56. O. Nakamura, Microsc. Res. Tech. 47, 165 (1999)57. A. Esposito, F. Federici, C. Usai, F. Cannone, G. Chirico, M. Collini, A. Di-

aspro, Microsc. Res. Tech. 63(1), 12 (2004)58. A. Diaspro, C.J.R. Sheppard, in Confocal and Two-Photon Microscopy: Foun-

dations, Applications and Advances, ed. by A. Diaspro (Wiley, New York,2002), p. 39

59. G.C. Cianci, J. Wu, K.M. Berland, Microsc. Res. Tech. 64(2), 135 (2004)60. C. Xu, J. Guild , W.W. Webb, W. Denk, Opt. Lett. 20, 2372 (1995)61. P.T.C. So, C.Y. Dong, B.R. Masters, K.M. Berland, Annu. Rev. Biomed. Eng.

2, 399 (2000)62. D.W. Piston, B.R. Masters, W.W. Webb, J. Microsc. 178, 20 (1995)63. M. Chalfie, Y. Tu, G. Euskirchen, W.W. Ward, D.C. Prasher, Science

263(5148), 802 (1994)64. S.M. Potter, C.M. Wang, P.A. Garrity, S.E. Fraser, Gene 173, 25 (1996)65. M. Zimmer, Chem. Rev. 102, 759 (2002)66. G.A. Blab, P.H.M. Lommerse, L. Cognet, G.S. Harms, T. Schmidt, Chem.

Phys. Lett. 350, 71 (2001)67. C.J.R. Sheppard, M. Gu, Optik 86, 104 (1990)68. O. Nakamura, Optik 93, 39 (1993)69. M. Gu, C.J.R. Sheppard, J. Microsc. 177, 128 (1995)70. G.H. Patterson, D.W. Piston, Ultramicroscopy 78, 2159 (2000)71. A. Diaspro, G. Chirico, C. Usai, P. Ramoino, J. Dobrucki, in Handbook of

Biological Confocal Microscopy, 3rd edn., ed. by J. Pawley (Plenum, New York,2006), p. 690

72. W. Denk, K.R. Delaney, A. Gelperin, D. Kleinfeld, B.W. Strowbridge, D.W.Tank, R. Yuste, J. Neurosci. Methods 54(2), 151 (1994)

73. A. Hopt, E. Neher, Biophys. J. 80, 2029 (2001)74. A. Abbotto, G. Baldini, L. Beverina, G. Chirico, M. Collini, L. D’Alfonso,

A. Diaspro, R. Magrassi, L. Nardo, G.A. Pagani, Biophys. Chem. 114(1), 35(2005)

75. T. Wilson, in Confocal and Two-Photon Microscopy: Foundations, Applicationsand Advances, ed. by A. Diaspro (Wiley, New York, 2002), p. 19

76. V.E. Centonze, J.G. White, Biophys. J. 75(4), 2015 (1998)77. K. Konig, H. Liang, M.W. Berns, B.J. Tromberg, Opt. Lett. 21, 1090 (1996)


78. K. Konig, I. Riemann, P. Fischer, K.J. Halbhuber, Cell. Mol. Biol. 45, 195(1999)

79. K. Konig, J. Microsc. 200, 83 (2000)80. K. Konig, T. Krasieva, E. Bauer, U. Fiedler, M.W. Berns, B.J. Tromberg, K.O.

Greulich, J. Biomed. Opt. 1, 217 (1996)81. P.E. Hanninen, S.W. Hell, Bioimaging 2, 117 (1994)82. J.B. Guild, C. Xu, W.W. Webb, App. Opt. 36, 397 (1997)83. R. Wolleschensky, T. Feurer, R. Sauerbrey, U. Simon, Appl. Phys. B 67, 87

(1998)84. R.H. Webb, Rep. Prog. Phys. 59, 427 (1996)85. A. Diaspro, G. Chirico, F. Federici, F. Cannone, S. Beretta, M. Robello, J.

Biomed. Opt. 6(3), 300 (2001)86. B.R. Masters, P.T. So, E. Gratton, Biophys. J. 72, 2405 (1997)87. L. Moreaux, O. Sandre, S. Charpak, M. Blanchard-Desce, J. Mertz, Biophys.

J. 80(3), 1568 (2001)88. P.J. Campagnola, M.D. Wei, A. Lewis, L.M. Loew, Biophys. J. 77(6), 3341

(1999)89. P.J. Campagnola, A.C. Millard, M. Terasaki, P.E. Hoppe, C.J. Malone, W.A.

Mohler, Biophys. J. 82(1), 493 (2002)90. A. Zoumi, X. Lu, G.S Kassab, B.J Tromberg, Biophys. J. 87(4), 2778 (2004)91. W.R. Zipfel, R.M. Williams, R. Christie, A.Y. Nikitin, B.T. Hyman, W.W.

Webb, Proc. Natl. Acad. Sci. USA 100(12), 7075 (2003)92. S. Rotha, I. Freund, J. Chem. Phys 70, 1637 (1979)

5

Applications of Optical Resonanceto Biological Sensing and Imaging: I. SpectralSelf-Interference Microscopy

M.S. Unlu, A. Yalc.in, M. Dogan, L. Moiseev, A. Swan, B.B. Goldberg,and C.R. Cantor

5.1 High-Resolution Fluorescence Imaging

Fluorescence microscopy is an essential tool in modern biological research.It is a powerful method that allows noninvasive monitoring of specificallylabeled targets within living cells, and simultaneous detection of multiple tar-gets using different labels. The spatial resolution in fluorescence microscopy islimited because of the diffraction limit; the resolution in transverse directionis proportional to λ/2NA = λ/2nsinθ (where n is the refractive index in theobject space, and θ is the half-angle of the largest cone of rays that can enteror leave the optical system), whereas the longitudinal resolution is given by2λn/NA2. High-spatial resolution to detect fluorescent molecules below thediffraction limit can be achieved in several ways, such as by increasing theeffective numerical aperture (as in 4Pi confocal microscopy) [1], introducingspatial variation in the excitation light creating finer spatial features in theimage (as in standing wave microscopy) [2], using multiple-photon fluores-cence absorption or emission mechanisms that lead to nonlinear effects in thelight field (as in 2-photon microscopy) [3], and by selectively quenching thefluorescence from a focal spot to obtain a very small fluorescing volume (asin stimulated emission depletion microscopy) [4].

5.2 Self-Interference Imaging

Fluorophores, when immobilized on surfaces, are often quenched because ofmechanisms of energy transfer, standing wave nodes in the excitation field,and destructive interference in the emission. Fromherz and coworkers notedthat the intensity of the total fluorescence oscillates as a function of the fluo-rophore height above a reflecting substrate [5]. Their technique, fluorescenceinterference contrast microscopy (FLIC), is based on measuring the intensityof fluorophores located within ∼λ of vertical distance from a reflecting sur-face. On the one hand, this proximity to the surface causes the entire emission

72 M.S. Unlu et al.

Fig. 5.1. The interferometric experimental configurations. (Left): WL reflectionspectroscopy is based on spectral variations of reflection from thin transparent films.Interference of light reflected from the top surface and a buried reference surfaceresults in periodic oscillations. (Right): The SSFM technique maps the spectraloscillations emitted by a fluorophore located on a layered reflecting surface into aprecise position determination.

spectrum to oscillate as light undergoes constructive and destructive interfer-ences. The fluorophore height can be determined from these oscillation curves;however, careful calibration of fluorescence intensity as a function of distancefrom the surface is required. On the other hand, for larger separations betweenthe fluorophore and the reflecting surface (on the order of 10–20λ), light cango between constructive and destructive interferences multiple times even atthe same height. This creates interference fringes in the emission spectrum,and spectral self-interference fluorescence microscopy (SSFM) [6–8] utilizesthis principle to reveal height information.

Using a combination of a traditional reflection technique and an interfer-ometric fluorescence spectroscopy technique, the average optical thickness ofbiological layers and the height of fluorescent markers can be measured withsubnanometer accuracy. Figure 5.1 summarizes the two techniques schemat-ically. The first technique, white light (WL) reflection spectroscopy, is basedon spectral variations of reflection from thin transparent films. Interference oflight reflected from the top surface and a buried reference surface results inperiodic oscillations in the reflection spectrum. The principle is similar to theinterference-based detection technique using color variations due to increasedpath length as a consequence of surface binding on optically coated silicon [9].The spectral measurements result in very high accuracy (∼0.2 nm) for measur-ing average optical density, comparable to ellipsometry, and provide a preciserelative measure of the additional “optical” mass on the substrate. The second

5 Spectral Self-Interference Microscopy 73

Fig. 5.2. Emission spectra of fluorescein immobilized on a glass slide and on top ofa Si-SiO2 chip with two different thicknesses of the oxide layer.

technique, SSFM, analyzes the spectral oscillations due to self-interference ofdirect and reflected emission by a fluorophore located on a layered reflect-ing surface and yields a precise position determination (Fig. 5.1). Figure 5.2illustrates the effect of self-interference on the emission spectrum of fluores-cein immobilized at different heights on a silicon–silicon oxide wafer. Thisis compared with the smooth emission envelope of fluorescein immobilizedon a glass slide where there is no self-interference. As shown, small heightdifferences (10 nm in the figure) shift the fringes and change the period ofoscillation, although the latter is less apparent.

It should be noted that in contrast to fluorescence interference-contrast mi-croscopy [5], the axial position of the fluorophores in SSFM is determined fromthe spectral oscillations and not from the total intensity; therefore, variationsin fluorophore density, emission intensity, and the excitation field strength donot affect the result.

5.3 Physical Model of SSFM

5.3.1 Classical Dipole Emission Model

An emitting fluorophore can be represented as a radiating dipole [7]. Theemitter transition dipole, µ, the wave vector, k, and the electric field vector, E,lie in the same plane, which is the plane of polarization of the radiated light.In the far field, if the environmental factors remain constant, the amplitude ofthe electric field of a fluorescently emitted wave is proportional to the sine ofthe angle (α) between the dipole µ and the wave vector k. The emission of aclassical dipole is, therefore, nonuniform, and has a donut shape (Fig. 5.3). Theemission pattern of a dipole near a surface can be described by consideringthe intensity and polarization of both the direct and the reflected waves.Consider that light emitted from a dipole with polar tilt angle θ placed at

74 M.S. Unlu et al.

E

kµ

a

Electric field

Direction ofpropagation

E a sina

Dipole

Fig. 5.3. Left : Intensity and polarization of electric field emitted by a classicaldipole. Right : 3D emission pattern of a classical electric dipole.

Plane ofincidence

Dielectricsurface

Plane ofincidence

z

kdtr

ETM

ETE

E

krefl

k1

q

j

µ

qerr

qe

ain

Fig. 5.4. Left : Direct and reflected waves emitted by a randomly oriented dipole.Right : Transverse electric and transverse magnetic components of the electric fieldof incident wave.

a distance d from a stack of mirrors is collected by a microscope objective,where the numerical aperture is defined by the collection cone of the objective(Fig. 5.4). Every emission direction within the collection cone can be describedby θem, the tilt angle from the surface normal, and ϕ, the azimuthal angle tothe dipole. The dipole radiates two coherent waves: one sent directly to thedetector, and one incident on the mirror surface, which is then reflected sothat it propagates parallel to the direct wave. As the distance between thedipole and the mirror is significantly smaller than the distance between thedipole and the microscope objective, these parallel waves arrive at the samespot on the detector.

The ETM component of both the incident and direct waves lies in theplane of incidence, and ETE is perpendicular to it. After reflection, the inci-dent wave is modified by reflection coefficients RTE and RTM of the stack ofmirrors. Also, there is an additional phase shift between the direct and the


reflected waves because of the distance the incident light travels from thesource to the stack of mirrors and back (ei2kd cos θem). Light intensity collectedby a microscope objective with maximum collection angle θmax

em is integratedover dθem sin θem. To account for fluorophores with different transition dipoleorientations, the total collected light intensity is integrated over all possibledipole orientations. Usually, fluorophores in monolayers are randomly distrib-uted in the horizontal plane, so their emission is integrated over ϕ; however,the range of polar tilt angles can sometimes be restricted.

For large enough separations between the fluorophores and the mirror,several oscillations of constructive and destructive interference appear withina small span of emitted wavelengths. This creates interference fringes in theemission spectrum, and the phase and contrast of the oscillations will dependon the location and orientation of dipoles above the mirror.

The emission of the dipole can be formulated as:

I = |ETE dir + ETE refl|2 + |ETM dir + ETM refl|2 , (5.1)

where

ETE dir = ETE inc ∝ cos θ sin ϕ, (5.2)

ETM dir/inc ∝√

1 − (sin θem sin θ cos ϕ ± cos θem cos θ)2 − sin2 θ sin2 ϕ, (5.3)

ETE refl = ETE incRTEei2φ, (5.4)

ETM refl = ETM incRTMei2φ, (5.5)

φ =4πn

λd cos θem, (5.6)

and n is the index of refraction of the medium surrounding the dipole.The total emission of a monolayer of random dipoles measured with an

objective of maximum collection angle θmaxem is:

Itotal =

π/2∫θ=0

π/2∫ϕ=0

θmaxem∫

θem=0

I (θ, ϕ, θem)dθdϕdθem sin θem. (5.7)

5.4 Acquisition and Data Processing

5.4.1 Microscope Setup

The measurements were performed with a system that combines an uprightLeica DM/LM microscope and a Renishaw 100B micro-Raman spectrometer.The sample was positioned and scanned using a motorized submicrometerprecision stage. An 1,800 grooves per millimeter grating was used with aspectral resolution of 2 cm−1 at 500 nm. The light dispersed by the grating

76 M.S. Unlu et al.

Optical path 488 nm notch filter

Spectrometer

Confocalaperture

Objective

Fluorophoremonolayer

Substrate

θmax

Laser488 nm

Fig. 5.5. Microscope setup for SSFM.

was imaged onto a thermoelectrically cooled charge-coupled device (CCD).For WL measurements, normal Koehler illumination with a halogen lamp wasused, whereas for fluorescence measurements, the light source was an argonion laser with a 488 nm line. The emission from the fluorophores was collectedwith a 5× (NA=0.12) objective and transmitted into the spectrometer througha notch filter, blocking the excitation light (Fig. 5.5).

5.4.2 Fitting Algorithm

Self-interference spectra are raw data composed of the envelope function rep-resented by the emission profile of the free fluorophore, the oscillatory interfer-ence component, and high-frequency Gaussian noise introduced by the spec-trophotometer. Both WL reflectivity and fluorescence self-interference spectrawere fitted using a custom-built MATLAB application that separates the os-cillatory component from the envelope function. This program automaticallycalculates the parameters of the system such as the thickness of thin filmsor position of the emitters above the mirror. The variation in the index ofrefraction of silicon oxide within the used wavelength span was taken into ac-count in the fitting algorithm. The model also takes into account the complexreflectivity of the underlying stack of dielectrics and the orientation of thedipole moments of the emitters. Because the curve-fitting algorithm extractsthe oscillatory term to determine the label height, the measurements are im-mune not only to potential quenching of the entire spectrum, but also to anyspectral modifications or nonradiative transfer effects.


5.5 Experimental Results

5.5.1 Monolayers of Fluorophores on Silicon Oxide Surfaces:Fluorescein, Quantum Dots, Lipid Films

SSFM can be used to measure the vertical position of fluorophores separatedfrom a silicon mirror by a transparent layer of silicon oxide. Monolayer surfacesof fluorescein, quantum dots [7], and lipid films [6] are investigated in theexperiments below.

Fluorescein

To demonstrate detection of the axial position of a fluorescein monolayerfrom the silicon mirror, a sample was prepared with a spacer layer of siliconoxide etched to different heights, and a monolayer of fluorescein was immo-bilized on surface via isothiocyanate-aminosilane chemistry. The surface wasthen scanned in a grid pattern with fluorescence emission spectra taken every200 µm, and the spectra were processed to yield the axial position of the fluo-rophores at each spot. Figure 5.6 displays the height data as a 3D gray-scaleimage, where it is apparent that nanometer axial resolution is obtained. Notethat this is not the fluorescence intensity, which is relatively uniform.

Quantum Dots

Quantum dots (QDs) are a new area of research with potential applicationsin many fields. The excitation spectrum of most QDs is unusually broad.However, the emission properties of a QD depend on its size: smaller dots emitlight with shorter wavelengths. Samples of QDs that are uniform in size have

Fig. 5.6. Surface profile reconstructed from self-interference spectra.

78 M.S. Unlu et al.

04.930

<1nm3nm

a)

b)

FluoresceinQuantum dots

4.935

4.940

4.945

4.950

4.955

2 4 6

chip.wl

QDs.wl

QDs.sfm

8 10Position on the chip, mm

Hei

ght (

µm)

12 14 16 18 20

Fig. 5.7. (a) Average position of fluorescein and quantum dots above a Si-SiO2

surface. (b) WL and SSFM measurements of a Si-SiO2 chip before and after quantumdot deposition. QDs are sparse enough that their effect on determining the spacerlayer thickness is insignificant. SSFM measurements show the source of emission isabout 3 nm above the surface.

sharp emission peaks; for a broader distribution of sizes, the spectra widen aswell. On the one hand, as a fluorescent label for biological research, quantumdots hold a number of advantages over conventionally used organic dyes asthey are photostable, can withstand many excitation-emission cycles, and canprovide a whole range of nonoverlapping spectra. On the other hand, QDsare less commonly used in staining biological specimens as they cannot be aseasily functionalized with reactive groups for selective attachment. Chemicalmethods of adapting QDs to new environments are being developed by manyresearch groups; in a recent achievement, QDs were used as fluorescent labelsinside living cells [10].

To demonstrate visualization of a monolayer of QDs immobilized on asurface, a silicon oxide chip was silanized with aminopropyltriethoxysilane(APTES). CdSe quantum dots capped with ZnS were treated with mercap-toacetic acid to render them hydrophilic and negatively charged at neutral pH.The QDs were then electrostatically attached to the aminated surface [11].SSFM measurements show that the average position of the emitters is ele-vated above the surface by about 3 nm (Fig. 5.7), which reflects the thicknessof the QDs themselves. When compared with a monolayer of fluorescein, thebest-fitting curve for the QD spectrum corresponds to a random orientation ofthe transition dipoles, whereas, in the case of fluorescein, the molecules seemto be more concentrated in the horizontal position.

Lipid Films

Deposition of lipid bilayers on surfaces has attracted considerable attentionbecause of the possibility of creating biomembrane-based biosensors and forstudying the fundamental properties of biomembranes. Artificial lipid filmscan be used as substitutes for biomembranes in investigating the structuraland functional properties of various transmembrane properties in conditionssimilar to those found in cells.


3.5 nm(fluorescence)

5.5 nm(white light)

4.938

4.936

4.934

4.932

4.930

4.928

0.0 2.0

Top LayerBottom Layer

4.0 6.0

Position on sample (mm)

Flu

orop

hore

Pos

ition

(µm

)

8.0 10.0

Fig. 5.8. Left : Structure of the lipid bilayer film. Right : Axial fluorophore positionsmeasured across the surface.

The interaction between the lipid films and various membrane-bound com-ponents can be probed by diagnostic tools such as infrared spectroscopy,neutron reflectivity, surface-plasmon resonance (SPR), and atomic force mi-croscopy (AFM). The finest resolution of lipid surface features can be achievedby AFM, but AFM is unsuitable for probing structures that are too delicateand cannot resolve buried components.

SSFM offers a unique opportunity to look inside the lipid membrane anddetect the position of a fluorescently labelled component. To determine theposition of a fluorescent label attached to the head groups of either the topor the bottom leaflet of a lipid bilayer separated by only about 4 nm, a layerof DPPE containing 2% DHPE-fluorescein was deposited by the Langmuir-Blodgett technique, followed by a layer of DPPE without the dye. The fluores-cence emission spectra as well as WL reflectivity measurements were taken atthe same locations on the chip. The chip was cleaned, followed by depositionof a new bilayer of lipids with the fluorescein-containing leaflet on top, andthe measurements were repeated. The results are summarized in Fig. 5.8. TheWL reflectivity measurements before and after lipid deposition were the samefor both experiments (not shown). The fluorescence emission spectra show theaverage position of the fluorescent dye attached to either the top or the bottomhead groups very accurately, fractions of a nanometer, inside the lipid bilayer.

5.5.2 Conformation of Surface-Immobilized DNA

DNA array technology, offering highly paralleled detection capability, has be-come a widespread tool in biological research with applications in expressionscreening, sequencing, and drug discovery. One of the defining characteristicsof a DNA array is the availability of the single-stranded probes for hybridiza-tion with the target. The conformation of DNA molecules in an array maysignificantly affect the efficiency of hybridization; immobilized molecules lo-cated farther away from the solid support are closer to the solution state, and

80 M.S. Unlu et al.

are more accessible to dissolved analytes. Recently, advances have been madeto characterize surface-bound DNA probes, using optical or contact methodssuch as ellipsometry, optical reflectivity [12,13], neutron reflectivity [14], X-rayphotoelectron spectroscopy [15], FRET [16,17], SPR [18,19], and AFM [20,21].However, most experimental techniques characterize the DNA layer as a sin-gle entity, parameterizing its thickness or density. The advantage of SSFM isthat unlike these other methods, it has the capability of examining the specificpositions of internal elements of the DNA chain.

To study the conformation of single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) on glass surfaces with SSFM, 50 and 21-nt oligonu-cleotides were used [8]. In all studies, the first strand of the DNA wascovalently bound to the surface at its 5′ end. Experiments were performedwith fluorescein markers bound to either the first strand at its distal 3′ end,or the second strand at its 3′ or 5′ end.

WL Reflectivity Measurements

As schematically shown in Fig. 5.9, oligonucleotides carrying a 5′-aminotag are covalently bound to an aminated surface via a homobifunctionalcrosslinker. Using WL reflectivity, progressive growth of the surface-boundthin films during DNA immobilization steps is determined. The thicknessof the silane layer is 0.8–1.0 nm, which roughly corresponds to a monolayer;phenylene isothiocyanate adds another 0.5–0.6 nm. Immobilization of DNAleads to a further increase in the optical thickness. The average film thicknessis determined by WL reflection spectroscopy assuming an index of refraction of1.46 for DNA as was measured for dense layers [22]. The optical film thicknessfor ssDNA of 21- and 50-nt is measured as 1.0–1.5 nm and 2.0–2.5 nm, respec-tively. Although the precision of absolute thickness will depend on the index,WL reflectivity provides an accurate relative measure of additional mass onthe surface and can thus monitor the efficiency of hybridization. As shown in

Fig. 5.9. The WL interference measurements of 50mers immobilized on the surfacebefore (left) and after (right) hybridization. The height of the transparent layers(2 and 3 nm for left and right, respectively) correspond to average optical thicknessof the DNA layer.


Fig. 5.10. The fluorescence interference measurements for the 50-bp dsDNA labeledat the proximal (left) and distal end (right).

Fig. 5.9 (right), adding complementary second strands to 50-mers results inan increase in the film thickness by ∼1.0 nm corresponding to a hybridizationefficiency of ∼50%.

Determination of the Position of a Fluorescent Label

The optical thicknesses of the layers obtained by WL reflectivity measure-ments can be used together with the SSFM measurements to determine theposition of the fluorophores relative to the surface. Figure 5.10 illustrates theschematics of the experiment to study the elevation of dsDNA fragments. Themaximum elevation of the label is limited by the length of the double helix,which is ∼7 nm for 21-bp and ∼17 nm for 50-bp fragments. As dsDNA has apersistence length of ∼50 nm [23], short fragments in the experiments can beviewed as rigid rods on hinges. A rigid rod hinged to the surface would havean average height of the distal label with an average tilt angle of 60◦ fromnormal, assuming free rotation of one half of the length. Factors such as sterichindrance or surface interactions may limit this free rotation. Using SSFM, wemeasured the elevation of the label on top of DNA double helices to be 5.5 nmfor 21-bp fragments and 10.5 nm for 50-bp fragments, which represents tiltangles from the normal of 40◦ and 50◦, respectively. These values are a mea-sure of the average distribution of heights within the microscope focal spot. Ifthe label on the second strand is at the 3′ end, its location in the double helixwill be at the bottom, close to the surface. Although we did not expect to seea significant variation, the proximal, 3′-end label on 50-bp DNA is elevatedby ∼2–3 nm compared with only 0.5–1 nm in the case of the 21-bp fragment.This is because a 50-mer is long enough to have stable partial hybridizationthat could free the proximal end and yield an elevated label position, and alsobecause of steric hindrance.

We also studied the conformation of ssDNA by measuring the height of flu-orescent tags attached to the free end of surfacebound DNA oligonucleotides.Unlike dsDNA, ssDNA is flexible and little is known about the shape or size ofssDNAs on the surface. AFM measurements suggest that ssDNA immobilizedon a surface exists in a globular conformation [21].

82 M.S. Unlu et al.

However, there are reports stating that because of steric hindrance fromnearby molecules, ssDNA may change its conformation from a random coil tomore extended forms [22,23]. The fluorescent label attached to the distal endof surface-bound single-stranded 21-mer is found to be close to the surface,within 1 nm. In a similar situation, 50-mers show much higher location of thelabel: 5.5 nm above the surface as illustrated, pointing out to a considerablymore extended conformation compared with 21-mers, which may be due tosteric effects from closely located grafts in the DNA layer. The surface den-sity of immobilized ssDNA measured using a radiolabel is ∼35 fmol/mm2 forboth 21- and 50-mers, which translates into 11 nm distances between adjacentmolecules or a 5.5 nm radius of free space around each. As the length of a fullyextended 50-mer is 27.5 nm, it is enough to interact with its neighbors, at leastintermittently. When a second, unlabeled strand is hybridized, the label at thedistal end of the newly formed duplex extends out as well. Unlike the dsDNAin Fig. 5.10, the DNA layer now consists of two species: the dsDNA and theunhybridized strands, both of which are carrying a fluorescent marker but atdifferent heights. The average position of the marker should be somewhere inbetween depending on the degree of hybridization. The binding efficiency isestimated by comparing the average height of the fluorescent layer above thesurface for the two cases, and found to be between 30% and 50%. This roughestimate is close to the result obtained by WL reflectivity: 50% hybridizationfor both 21 and 50-mers, demonstrating self-consistency. As a further check,we have performed density measurements with radiolabeled DNA and foundit consistent with 50% hybridization. Because of intrinsic limitations such assubstrate-related quenching of radiative emission, we do not consider the ra-diolabeling for absolute determination of DNA densities, but rather only forrelative estimation of DNA densities.

We also studied how the conformation of a surface-bound 50-nt labeledoligonucleotide changes when it is annealed with a 21-mer complementaryto either its top or bottom part. When a 21-mer complementary to the topsection was annealed, the position of the distal end increased from 5.5 to6.5 nm. However, when an ssDNA–dsDNA construct has the doublestrandedpart proximal to the surface, the position of the label is lower than that ofan unhybridized oligo, decreasing in average height from 5.5 to 3.5 nm. Thesedata suggest that the distal bound 21-mer construct is nearly vertical, and theproximal bound construct has formed a rotation point allowing the flexibledistal end to approach closer to the surface. The selected sequence of theoligonucleotides rules out the possibility of intramolecular DNA structures.

5.6 SSFM in 4Pi Configuration

SSFM is a powerful tool in determining the position of fluorescent labelsand giving insight about the conformation of biological structures boundto surfaces. However, despite its high precision axial position determination


capability, its lateral resolution is limited by several micrometers as low NAobjectives have to be used to ensure spectral fringe visibility. The large spacerlayer thickness results in a focal mismatch between the direct and the reflectedimage of the sample when high NA objectives are used. Also, with high NAobjectives, the rapid change of fields for large angles due to the interferenceof direct and reflected signals smears out the spectral interference fringes andlowers the signal contrast. One remedy for increasing lateral resolution byusing high NA objectives while maintaining the nanometer level axial local-ization capability is to use a second objective instead of a mirror and collectthe emission from both sides (Fig. 5.11a). As the spherical wavefronts of theemission are collected by a symmetrical configuration, path length mismatchfor higher collection angles for two interference pathways is eliminated unlikeusing one objective and a reflecting substrate. The path lengths of the twointerference arms are adjusted by one of the mirrors mounted on a piezo con-troller, such that they are different by several tens of micrometers to inducemodulations in the spectrum of the emitter. This configuration is similar to

a) b)

c)

1

4

3.5

3

2.5

2

1.5

1

0.5

0.8

0.6

0.4

0.2

018500

28 30 32 34

5.1 nm

36 38

18750 19000wavenumber [1/cm]

axial position [nm]

norm

aliz

ed in

tens

ity

Err

or [a

.u.]

19250 19500

Fig. 5.11. (a) A 4Pi microscope. (b) Normalized spectra of a monolayer of AlexaFluor 488 measured using 4Pi microscope at two axial positions that are 5 nm apart.(c) Minimization of error in fitting spectral responses of two positions.

84 M.S. Unlu et al.

4Pi confocal fluorescence microscopy introduced by Stefan Hell [1]. However,instead of using the 4Pi microscope for 3D image scanning, the emitted signalis collected by two high NA objectives and coupled to a spectroscopy systemto localize monolayers with nanometer accuracy. Scanning of the tight collec-tion spot is performed in lateral dimensions and axial profile is extracted formthe spectral fringes similar to SSFM.

In Fig. 5.11b, typical spectral response of a 4Pi SSFM system is shown. Thetwo distinct spectra are from the same spot of a monolayer that was moved by5 nm in axial direction between the data acquisition. The fluorescent emitteris a monolayer of Alexa Fluor 488 dye deposited on a glass cover slip andplaced in the common foci of the two high NA oil immersion objectives ofthe 4Pi microscope. The sample was excited by a 488 nm continuous wavelaser using the two objectives. The spectral response clearly shows the shift inspectral fringes because of the change in the axial position of the monolayer offluorophores. A simple fitting algorithm that utilizes a geometrical model andonly includes the position of monolayer with respect to the geometrical focusand the external path length difference can be used to determine the relativedifference in the axial positions of monolayers. As seen from Fig. 5.11c, theaxial position values that minimize the error in fitting the spectral responsesof two positions of the monolayer were actually 5.1 nm apart, which agreeswell with the 5 nm physical change in the axial position of the layer.

5.7 Conclusions

We have developed a new technique, SSFM, which transforms the variation inemission intensity for different path lengths used in fluorescence interferome-try to a variation in the intensity for different wavelengths in emission, encod-ing the high-resolution information in the emission spectrum. Using SSFM,we have demonstrated analysis of systems that include monolayers of fluores-cein, quantum dots, and lipid films. More importantly, we have demonstratedconformation of surface-immobilized DNA by estimating the shape of coiledssDNA, the average tilt of dsDNA of different lengths, and the amount ofhybridization. We have also shown that localization of axial position of fluo-rescent structures with high precision is possible without sacrificing the lateralresolution using SSFM in 4Pi configuration.

These data provide important proofs of concept for the capabilities ofnovel optical methods for analyzing the molecular disposition of DNA andprotein on surfaces. The determination of DNA conformations on surfaces andhybridization behavior provide information required to move DNA interfacialapplications forward and thus impact emerging clinical and biotechnologicalfields.


Acknowledgments

The authors acknowledge Prof. Irs.adi Aksun of Koc University for hiscontributions to modeling of fluorophore emission on layered surfaces andDr. Stephen Ippolito for his help in setting up the microscope, the NationalScience Foundation (Grant DBI0138425), Air Force Office of Scientific Re-search (Grant MURI F-49620-03-1-0379), and National Institutes of Health–National Institute of Biomedical Imaging and BioEngineering (Grant 5R01EB00 756-03) for their support. The views and conclusions contained inthis document are those of the authors and should not be interpreted asrepresenting the US Government.

References

1. S. Hell, E.H.K Stelzer, JOSA A 9(12), 2159 (1992)2. B. Bailey, D.L. Farkas, D.L. Taylor, F. Lanni, Nature 366, 44 (1993)3. P.T.C. So, C.Y. Dong, B.R. Masters, K.M. Berland, Ann. Rev. Biomed. Eng. 2,

399 (2000)4. M. Dyba, S.W. Hell, Phys. Rev. Lett. 88, 163901 (2002)5. A. Lambacher, P. Fromherz, Appl. Phys. A 63, 207 (1996)6. A.K. Swan, L. Moiseev, C.R. Cantor, B. Davis, S.B. Ippolito, W.C. Karl, B.B.

Goldberg, M.S. Unlu, IEEE J. Select. Top. Quantum Electron. 9, 294 (1996)7. L. Moiseev, C.R. Cantor, I. Aksun, M. Dogan, B.B. Goldberg, A.K. Swan, M.S.

Unlu, J. Appl. Phys. 96, 5311 (2004)8. L. Moiseev, M.S. Unlu, A.K. Swan, B.B. Goldberg, C.R. Cantor, Proc. Natl.

Acad. Sci. 103, 2623 (2006)9. R. Moller, A. Csaki, J.M. Kohler, W. Fritzsche, Nucleic Acids Res. 28, e91

(2000)10. X. Wu, H. Liu, J. Liu, K.H. Haley, J.A. Treadway, J.P. Larson, N. Ge, F. Peale,

M.P. Bruchez, Nat. Biotech. 21(1), 41 (2003)11. W.C.W. Chan, D.J. Maxwell, X.H. Gao, R.E. Bailey, M.Y. Han, S.M. Nie, Curr.

Opin. Biotech. 13(1), 40 (2001)12. L.A. Chrisey, G.U. Lee, C.E. O’Ferrall, Nucleic Acids Res. 24, 3031 (1996)13. D.E. Gray, S.C. Case-Green, T.S. Fell, P.J. Dobson, E.M. Southern, Langmuir 13,

2833 (1997)14. R. Levicky, T.M. Herne, M.J. Tarlov, S.K. Satija, J. Am. Chem. Soc. 120, 9787

(1998)15. T.M. Herne, M.J. Tarlov, J. Am. Chem. Soc. 119, 8916 (1997)16. M.T. Charreyre, O. Tcherkasskaya, M.A. Winnik, Langmuir 13, 3103 (1997)17. M.C. Murphy, I. Rasnik, W. Cheng, T.M. Lohman, T. Ha, Biophys. J. 86, 2530

(2004)18. K.A. Peterlinz, R.M. Georgiadis, T.M. Herne, M.J. Tarlov, J. Am. Chem. Soc.

119, 3401 (1997)19. L.K. Wolf, Y. Gao, R.M. Georgiadis, Langmuir 20, 3357 (2004)20. S.O. Kelley, J.K. Barton, N.M. Jackson, L.D. McPherson, A.B. Potter,

E.M. Spain, M.J. Allen, M.G. Hill, Langmuir 14, 6781 (1998)

86 M.S. Unlu et al.

21. L.S. Shlyakhtenko, A.A. Gall, J.J. Weimer, D.D. Hawn, Y.L. Lyubchenko,Biophys. J. 77, 568 (1999)

22. Q. Du, M. Vologodskaia, H. Kuhn, M. Frank-Kamenetskii, A. Volodogski,Biophys. J. 88, 4137 (2005)

23. S. Elhadj, G. Singh, R.F. Saraf, Langmuir 20, 5539 (2004)

6

Applications of Optical Resonanceto Biological Sensing and Imaging:II. Resonant Cavity Biosensors

M.S. Unlu, E. Ozkumur, D.A. Bergstein, A. Yalc.in, M.F. Ruane,and B.B. Goldberg

6.1 Multianalyte Sensing

Interrogating binding interactions between proteins, segments of DNA orRNA, and biospecific small molecules is critically important for a great num-ber of applications in biological research and medicine. Microarray technologyhas emerged over the past decade to address applications that seek to mea-sure thousands or even millions of binding interactions at once. A microarrayconsists of a solid support, or substrate, with multiplicity of spots on its topsurface each containing a different type of fixed capturing molecule. A samplesolution containing unknown target molecules, or analytes, is introduced tothe microarray surface typically via a small fluid chamber or flow cell. Theamount of target material bound to any feature after washing the array givesan indication of the affinity between the target and capturing agent at thatspot. There are a number of important applications in biological researchthat benefit from microarray throughput such as gene expression profiling orantigen–antibody interaction monitoring. Aside from research applications,microarrays may play a crucial role in a new era of medical diagnostics. Bio-medical research is continuing to point toward molecular biomarkers that canbe used to help doctors diagnose diseases sooner, with greater accuracy, andprovide information that helps doctors personalize treatment plans.

Present microarray technology requires that the target molecules in thesample solution be labeled or attached with a fluorescent dye molecule forthe purpose of determining how much target material has bound to each mi-croarray feature using a fluorescence detector. The need to label the targetmolecules to visualize the results is a significant shortcoming of present tech-nology and has been described previously [1–4]. In general, fluorescent labelingmay suffer from bleaching of the labels, quenching effects from the surface,or low contrast between the label and the autofluorescence of the microarraysubstrate. In addition, proteins in particular may suffer from altered bindingproperties once they are labeled. For these reasons, label-free detection for

88 M.S. Unlu et al.

both DNA and protein is preferred, particularly for use in medical diagnosticswhere variation and cost must be minimized.

A vast number of label-free detection techniques have surfaced over thelast decade or two to address the problems of label-based detection [5–16].The important figures of merit for these techniques are sensitivity to boundtargets, throughput (number of features evaluated in a single test) and theability to scale throughput, and cost per test. Sensitivity is often quantifiedby mass per area (pg mm−2) or average height (pm) where 1 pm correspondsto roughly 1 pg mm−2 for protein or DNA [17–25].

6.2 Resonant Cavity Imaging Biosensor

6.2.1 Detection Principle

Two partially reflecting substrates are positioned such that their reflectingsurfaces face each other and form the optical cavity (Fig. 6.1) [26]. The tun-able wavelength laser light is collimated and incident from the back of oneof the reflectors. The wavelength of the laser is swept in time and at specificwavelengths the resonant condition of the cavity is met, the light resonatesinside the cavity and couples out. Beyond the cavity, the transmitted lightis imaged on a camera, so that the resonant response at each location of thecavity is recorded by a corresponding pixel on the camera. The probe biomole-cules are patterned on one of the reflector surfaces. When target biomoleculesbind to their specific capturing agents, the local resonant response shifts in

Fig. 6.1. Resonant cavity imaging biosensor (RCIB) setup.

6 Resonant Cavity Biosensors 89

wavelength. The presence of biomolecules bound to the surface is fundamen-tally detected by a small perturbation of the electric field inside the cavity.The effect of a slight phase delay is amplified by the resonant behavior of lightwithin the cavity. The cavity enhancement can provide a significant sensitiv-ity improvement over interference techniques that do not benefit from a highfinesse optical cavity. The reflectors can be constructed with alternating di-electric layers of silicon and SiO2 and may end with a layer of SiO2 that allowsthe surface immobilization chemistries that are developed for glass slides tobe used. The use of a parallel optical cavity, tunable wavelength laser, and adigital camera allows high throughput. Aside from the alignment and stabilityof the reflectors, and the internal operation of the tunable laser, there are nomoving parts in the setup. Additionally, the reflectors described here may bemade very inexpensive and hence disposable. In the present implementation,the cavity is formed with reflectors fabricated from alternating layers of Siand SiO2, constituting a Bragg grating (Fig. 6.2) [27], and the wavelength oflight is varied between 1,510 and 1,515 nm. The electric field distribution inthe cavity forms a standing wave pattern with a minimum at the Si surface ofthe reflectors. It is beneficial to place the sensing surface on an approximatelyquarter wavelength thick layer of similar low-index material, namely SiO2.Adding this extra oxide layer brings the binding surface some distance intothe cavity where the field is maximum (Fig. 6.3). When it is placed at the fieldmaximum, the sensing layer interacts with the highest number of photons andslight thickness changes will have a maximum effect on the field distribution,thereby shifting the resonant wavelength a maximum amount. If the sensingsurface is positioned at a field minimum, small changes on the surface will haveno effect on the resonant cavity and will therefore be undetectable. Addingan extra layer of SiO2 also has the previously mentioned benefit of providing

−1µm

Si

Si Substrate

SiO2

SiO2

Si

Fig. 6.2. The Bragg structures that are used as the reflectors in RCIB setup withalternating layers of 1/4 wavelength thick Si and 3/4 wavelength thick SiO2. Theimage is taken with SEM.

90 M.S. Unlu et al.

Fig. 6.3. Electric field pattern in the optical cavity.

Fig. 6.4. Simulation of the transmittance of the cavity. The reflectors are 40.0 µmapart from each other with air in between. A final SiO2 layer is included and shiftcorresponds to 5 nm step on the oxide.

a chemically compatible surface for microarray fabrication. Simulation of thecavity transmittance vs. wavelength with changing top surface oxide thicknessis plotted in Fig. 6.4. Simulation was done in Matlab using the matrix methodfor calculating transmittance through layered media [28].

6.2.2 Experimental Setup, Data Acquisition, and Processing

An RCIB setup has been constructed. The illumination source is an IR tunablelaser with a fiber-coupled output. Light is coupled into the system with a SMF-28 fiber and collimated to 10 mm FWHM. All the optics are antireflectioncoated to avoid interference effects. The laser power is set to 15 µW, and thewavelength is continuously swept from 1,510 to 1,515 nm in 10 s during datacollection. The reflectors, which are 15 mm by 15 mm, are placed 40 µm apart


from each other. One reflector is mounted on a tip-tilt stage to bring thereflectors parallel. The second reflector is mounted on a translational stage toadjust the reflector separation. The transmission from the cavity is imagedon an indium–gallium–arsenide (InGaAs) sensor array with 5 × magnificationand an aperture setting the NA to 0.1. With these settings, 12 × 12 µm2 areasare imaged on the 60 × 60 µm2 pixels of the sensor array, which has 128 ×128 pixels. The laser wavelength is swept at a rate of 0.5 nm s−1 while thecamera captures images at a rate of 30 fps. Images are digitized to 12 bits(4,096 gray-levels) and transferred to a PC.

These data are transferred to Matlab and fit using several different meth-ods. There is a trade-off between the accuracy and the speed with which thedata can be fit. A more accurate (noise-resilient) curve fitting method is to fitthe data to resonant line shape given by (6.1), which models a simple resonantcavity where

T =ni

nf

∣∣∣∣ t2ejkd

1 − r2e−2jkd

∣∣∣∣2

, (6.1)

ni and nf are the refractive indices in and beyond the cavity, t and r arethe transmission and reflection coefficients, respectively, of the reflectors, k isthe wavenumber inside the cavity, and d is the reflector separation (Fig. 6.5).Further image processing subtracts the curvature inherent to the reflectors.Finally, a surface profile of small height variations on the SiO2 surface can beobtained.

6.2.3 Experimental Results

The system sensitivity was tested by imaging etched features on the oxidesurface (Fig. 6.6). The reflector that has a SiO2 layer on top is used as the

Fig. 6.5. Resonant curve recorded and fitting to this data using (6.1).

92 M.S. Unlu et al.

Fig. 6.6. Left : RCIB scan of the overlapping boxes. Right : Line cut of the image intwo consecutive measurements.

sensing surface. Standard photolithography and wet (BOE) etching were usedto pattern these features on the surface. Two overlapping rectangles, withdepths of 3 and 6 nm were etched resulting in four regions with differentoxide thickness; not etched, 3 nm deep, 6 nm deep, and 9 nm deep etched.These values are consistent with the AFM measurements of the same sample.System repeatability was tested by making consecutive measurements of thesame sample. Averaging boxes measuring 5 pixels by 5 pixels, or an area of50 × 50 µm2, the RMS deviation in consecutive measurements was 0.01 nm(Fig. 6.6).

6.2.4 Spectral Reflectivity Imaging Biosensor

The phase delay added by extra material can also be imaged using a sim-pler approach. We propose a surface profilometry technique to be used asa label-free microarray imaging device. Spectral reflectivity imaging biosen-sor (SRIB) uses wavelength dependent reflectivity of a silicon substrate withthick thermally grown SiO2(5–10 µm), to accurately find the film thickness attens of thousands of different spots, and thus image the surface profile. Thistechnique, having a much lower finesse, is expected to be less sensitive thanRCIB, but should be more robust and offer higher throughput. A collimatedlaser beam that can be tuned from 764 to 784 nm is reflected from a pelliclebeam splitter and is incident on the sample surface (Fig. 6.7). At any fixedwavelength, reflection from the sample is imaged on a CCD camera (cam-era pixel size is 13.7 µm and the magnification of the imaging system is 2.3).Then the wavelength is stepped, and another image is taken. Repeating thisthrough the sweeping range of the laser, one can form a spectral reflectivitycurve for every 36 µm2 area on the sample in the field of view. The separate


Fig. 6.7. Spectral reflectance imaging biosensor diagram.

reflections from oxide and silicon surfaces form an interference pattern andan exact oxide thickness corresponding to each pixel is found by curve fittingthe recorded data.

Reflection of a single layer can be well estimated by Airy’s formulas [28].The reflection coefficient for an incoming field is given by:

r =r12 + r23e

−2iφ

1 + r12r23e−2iφ(6.2)

φ =2πd

λnSiO2 cos θ, (6.3)

where r12 and r23 are the Fresnel reflection coefficients from the SiO2 surfaceand SiO2-Si interface, respectively, and φ is the optical phase difference be-tween the two reflections. Here, d is the SiO2 thickness, nSiO2 is the refractiveindex of SiO2, λ is the wavelength of the incident light, and θ is the incidenceangle to the SiO2–Si interface, which will be 0◦ for perpendicular incidence.Curve fitting tools are used to fit this equation to the recorded data at eachpixel and to find the d. The setup was tested with etch marks of overlappingrectangles that was mentioned in the previous section. Silicon wafers with6.1 µm of thermally grown SiO2 were etched by photolithographic techniques.Three different depths of etch steps were created on the sample, which were4, 12, and 16 nm, and their depths were confirmed with AFM scans. The sam-ple was then imaged by SRIB, and data were processed with curve fitting. Aline-cut from the image can be seen in Fig. 6.8. These etch marks were imagedwith an accuracy of 0.5 nm.

94 M.S. Unlu et al.

06114

6116

6118

6120

6122

6124

6126

6128

6130

6132

6134

50 100pixels

dete

cted

oxi

de th

ickn

ess

(nm

)

150

Fig. 6.8. Line-cut from a processed image taken by the SRIB setup. Etch marksat three different heights are seen: 4, 12, and 16 nm. The thickness of the oxide was∼6,130 nm, initially.

Although the present data show that the system can image 250 spots withsubnanometer sensitivity, balancing the laser intensity fluctuations and usinga better imaging system is expected to improve sensitivity to 0.1 nm withmore than 10,000 spots.

6.3 Optical Sensing of Biomolecules Using MicroringResonators

6.3.1 Basics on Microring Resonators

Integrated devices featuring resonant microcavities with high quality factors(Q) such as toroids, disks, rings, and spheres have been used as add-dropfilters, optical switches, and in laser applications [29–33]. These devices haverecently become popular for research in biochemical sensing [34–38] as thedemand for highly sensitive and compact devices to detect biomolecules in-creased. Light is confined within the microring cavities by total internal re-flections resulting in high Q resonant modes. When the refractive index of thecladding or the outside medium changes (e.g., due to binding of molecules), anew guiding condition is obtained for the mode, causing a shift in the resonantwavelength.


Microring resonators provide sensitivities comparable to surface plasmonresonance because of their high Q (∼12,000 in this study) [16,39], and they arerobust and can be mass-manufactured using well-established silicon-integratedcircuit fabrication techniques. However, they lack the very high-throughputpotential that is demonstrated with RCIB.

The sensitivity of the system presented here is quantified through mea-surement of resonance shifts induced by a change in the refractive index ofthe medium (a bulk change). Additionally, sensing of biomolecules is demon-strated through observation of the change in resonant condition caused bymolecular binding of the well-documented avidin–biotin couple on the surface.

6.3.2 Setup and Data Acquisition

The microring resonators used in this experiment are 60 µm in radius, andvertically coupled to waveguides (Fig. 6.9). The effective index of the cavityis measured as neff ≈ 1.5 in deionized water (DI-H2O) ambience, and thepenetration depth for the guided mode at 1,550 nm in this environment iscalculated as ∼360 nm. A fiber-coupled tunable wavelength IR laser is usedas the light source. The cavity can support both TE and TM modes andwith the use of a polarizer one of the modes is selected. An optical splitterplaced after the polarizer separates the signal into two paths: one arm ischopped in free space at a frequency of 220 Hz and then coupled to the inputwaveguide of the microring resonator. The output waveguide is coupled to aphotodetector as signal input. The second arm acts as the reference input tothe photodetector for balanced detection. Common-mode noise cancellation

lateral offset

vertical offset

Input/output waveguides

ring resonator

out in

Fig. 6.9. Microring resonator is vertically coupled to input/output waveguides. Lat-eral and vertical offsets of the resonator allow control over the coupling coefficient.Receptor molecules are immobilized on the resonator surface, and target moleculesare released on the surface in a solution during flow. The top view of the resonatoris shown in the inset.

96 M.S. Unlu et al.

(mainly laser intensity fluctuations) is achieved through balanced detection,improving the signal-to-noise ratio, thus the overall system sensitivity. A flow-cell is used for controlled solution flow over the microring surface, and real timesignal during solution flow is acquired and analyzed with LabVIEW. Directmeasurement of a shift in resonance through repeated spectral scans is timeconsuming. Alternatively, the resonance shift can be measured indirectly bycollecting data at a single wavelength and observing the intensity in real time,and the change in intensity can then be mapped to a shift in resonance. Toobtain the maximum change in intensity due to a change in effective index, thewavelength at which the measurements are performed is selected to correspondto the point of maximum slope of the resonance curve. The high intensitystability necessary for these measurements is achieved by eliminating the noisecomponents of laser output through balanced detection.

6.3.3 Data Analysis and Discussion

The bulk experiments to determine the sensitivity of the system were con-ducted by flow of a solution of known refractive index with respect to therefractive index of DI-H2O that is used as a calibration reference. In Fig. 6.10,the measured shift in resonance with respect to a change in the refractiveindex of the medium (corresponding to various concentrations of phosphatebuffered saline (PBS) solution) is plotted. Notice that the relation is highly lin-ear. Using standard deviation δ (pm) of signal levels during PBS flow and theslope (m) of the plot in Fig. 6.10, the limit of detection (LOD) for a change inrefractive index can be approximated as LOD(n)=3δ m−1. We conclude thatLOD(n) = 1.8 × 10−5 refractive index units (RIU) for our setup. The experi-ment to demonstrate detection of biomolecules consists of two parts: bindingof biotinylated lectins to an avidin covered surface, and breaking the avidin–biotin bonds for surface regeneration. For binding, 3.5 µM biotinylated lectinsolution is flowed over the surface. Once the signal level reached by the bindingof biotin molecules to avidin covered surface is stabilized, DI-H2O is flowed towash away the unbound molecules. For surface regeneration, the surface is firstwashed with a chemical pH-7 buffer that breaks the avidin–biotin bonds, andthen with DI-H2O. The real-time intensity recording of binding/regenerationprocesses is shown in Fig. 6.11. The actual amount of change in the signal ismeasured between levels of DI-H2O flow before and after the introduction ofbiotinylated lectin solution. At these levels, the outside medium is identical,so the change in intensity is caused only due to binding of molecules. Thesignal level at the end of the regeneration phase is close to the initial levelindicating that partial recovery of surface is achieved.


1.3330

0

20

40

60

80

100

120

140

1.3332 1.3334

y = 141332x-188397R2=0.99458

Res

on

ance

Sh

ift

(pm

)

Refractive Index

1.3336 1.3338

Fig. 6.10. Shift of resonance is linearly related to change in refractive index. Theslope of the plotted curve is used to determine the limit of detection for refractiveindex changes.

BLBinding

DI-H2O

DI-H2O

Intensitychange

Break due todata acq.

pH

0

0.00

0.02

0.04

0.06

0.08

0.10

0.12

2000 4000 6000

Time (s)

Inte

nsi

ty (

V(r

ms)

)

8000 10000

Regeneration

Fig. 6.11. The detected intensity in a real time binding experiment is plotted. Theplot illustrates both binding and regeneration phases (BL: biotinylated lectin solu-tion, DI-H2O: deionized water, pH: pH-7 buffer). Binding efficacy is only measuredbetween the signal levels when DI-H2O is flowed across the sensor.

98 M.S. Unlu et al.

6.4 Conclusions

We have presented three optical techniques aimed at molecular sensing thatemploy optical resonance. A planar optical resonant cavity is used in thecase of RCIB, where the high finesse of the optical cavity can provide highsensitivity, while the planar configuration tied with a camera and tunablelaser can provide for simultaneous detection from many parallel locations. Thetechnique was demonstrated using nanometer scale etches in a SiO2 surfacethat model microarray features. We have also presented a second techniquethat observes the wavelength-dependent reflectivity from a silicon substratewith a thermally grown oxide layer to detect molecules binding to the top oxidesurface. In this case, the Fresnel reflection measured can be considered to bethe result of a low finesse optical cavity formed by the oxide layer together withthe bound molecules. The benefit of this technique compared with the highfinesse RCIB technique is that the monolithic structure avoids the difficultymaintaining the reflector alignment necessary in the latter technique. In thefinal technique presented, microring resonators employ a very strong resonantcoupling to achieve very high sensitivity at the cost of added complexity anda larger minimum feature size.

Acknowledgments

The authors acknowledge Boston University Office of Technology Develop-ment for their support, the National Institute of General and Medical Sci-ences (NIH R21 GM 074872-02), and the Army Research Laboratory (ARL)for their support under the ARL Cooperative Agreement Number DAAD17-99-2-0070 and under the AMCAC-RTP Cooperative Agreement DAAD19-00-2-0004. The authors also thank Dr. Brent Little and the Little Optics Divisionof Nomadics Inc. for providing the microring resonator devices. The views andconclusions contained in this document are those of the authors and shouldnot be interpreted as representing the Laboratory or the US Government.

References

1. S. Niu, R.F. Saraf, Smart Mater. Struct. 11, 778 (2002)2. R. Nadon, J. Shoemaker, Trends Genet.18, 265 (2002)3. M.A. Cooper, Nat. Rev. Drug Discovery 1, 515 (2002)4. M. Schena, Microarray Biochip Technology, (Eaton Publishing, Natick, 2000)5. J.M. Brockman, B.P. Nelson, R.M. Corn, Annu. Rev. Phys. Chem. 51, 41 (2000)6. K. Usui-Aoki, K. Shimada, M. Nagano, M. Kawai, H. Koga, Proteomics 5, 2396

(2004)7. C. Worth, B.B. Goldberg, M.F. Ruane, M.S. Unlu, IEEE J. Select. Top. Quan-

tum Electron. 7, 874 (2001)8. W. Lukosz, Biosensors Bioelectron. 6, 215 (1991)9. H. Nygren, T. Sandstrom, M. Stenberg, J. Immunol. Meth. 59, 145 (1983)


10. T. Sandstrom, M. Stenberg, H. Nygren, Appl. Opt. 24, 472 (1985)11. J. Piehler, A. Brecht, G. Gauglitz, Anal. Chem. 68, 139 (1996)12. P.I. Nikitin, M.V. Valeiko, B.G. Gorshkov, Sensors Actuators B 90, 46 (2003)13. J.P. Landry, X.D. Zhu, J.P. Gregg, Opt. Lett. 29, 581 (2004)14. J. Lu, C.M. Strohsahl, B.L. Miller, L.J. Rotherberg, Anal. Chem. 76, 4416

(2004)15. S. Chan, Y. Li, L.J. Rothberg, B.L. Miller, P.M. Fauchet, Mater. Sci. Eng. C

15, 277 (2001)16. A. Yalcin, K.C. Popat, J.C. Aldridge, T.A. Desai, J. Hryniewicz, N. Chbouki,

B.E. Little, O. King, V. Van, S. Chu, D. Gill, M.F. Anthes-Washburn, M.S.Unlu, B.B. Goldberg, IEEE J. Select. Top. Quantum Electron. 12, 148 (2006)

17. L. Moiseev, M.S. Unlu, A.K. Swan, B.B. Goldberg, C.R. Cantor, Proc. Natl.Acad. Sci. 103, 2623 (2006)

18. L.K. Wolf, Y. Gao, R.M. Georgiadis, Langmuir 20, 3357 (2004)19. L. Moiseev, C.R. Cantor, A.K. Swan, B.B. Goldberg, M.S. Unlu, Proc. Int. Soc.

Opt. Eng. Nanobiophotonics Biomed. Appl. 5331, (2004)20. J.J. Ramsden, Opt. Biosens. J. Mol. Recogn. 10, 109 (1997)21. J.A. DeFeijter, J. Benjamins, F.A. Veer, Biopolymers 17, 1759 (1978)22. S.N. Timasheff, in Handbook of Biochemistry and Molecular Biology, vol. 3, ed.

by G. Fasman (CRC Press, Cleveland, 1977), p. 37223. P. Schaaf, P. Dejardin, A. Schmitt, Langmuir 3, 1131 (1987)24. T.M. Davis, W.D. Wilson, Anal. Biochem. 284, 348 (2000)25. S. Nui, A. Label-free, Ph.D. dissertation, Virginia Polytechnic Institute and

State University, 200426. D. Bergstein, M.F. Ruane, M.S. Unlu, in International Semiconductor Device

Research Symposium, Maryland, USA, 7–9 December 200527. M.K. Emsley, O.I. Dosunmu, M.S. Unlu, IEEE J. Select. Top. Quantum Elec-

tron. 8, 948 (2002)28. P. Yeh, Optical Waves in Layered Media (Wiley, New York, 1988)29. B.E. Little, S.T. Chu, W. Pan, D. Ripin, T. Kaneko, Y. Kokubun, E.P. Ippen,

IEEE Photon. Technol. Lett. 11, 215 (1999)30. M. Cai, G. Hunziker, K. Vahala, IEEE Photon. Technol. Lett. 11, 686 (1999)31. V. Van, T.A. Ibrahim, K. Ritter, P.P. Absil, F.G. Johnson, R. Grover,

J. Goldhar, P.T. Ho, IEEE Photon. Technol. Lett. 14, 74 (2002)32. L. Yang, D.K. Armani, K.J. Vahala, Appl. Phys. Lett. 83, 825 (2003)33. B. Liu, A. Shakouri, J.E. Bowers, Appl. Phys. Lett. 79, 3561 (2001)34. K.J. Vahala, Nature 424, 839 (2003)35. R.W. Boyd, J.E. Heebner, Appl. Opt. 40, 5742 (2001)36. S. Arnold, M. Khoshsima, I. Teraoka, S. Holler, F. Vollmer, Opt. Lett. 28, 272

(2003)37. F. Vollmer, D. Braun, A. Libchaber, M. Khoshsima, I. Teraoka, S. Arnold, Appl.

Phys. Lett. 80, 4057 (2002)38. J. Hryniewicz, N. Chbouki, B.E. Little, O. King, V. Van, S. Chu, D. Gill, in

Biophotonics/Optical Interconnects and VLSI Photonics/WBM Microcavities,2004 Digest of the LEOS Summer Topical Meetings, 33 (2004)

39. Bardin, I. Kasik, A. Trouillet, V. Matejec, H. Gagnaire, M. Chomat, Appl. Opt.41, 2514 (2002)

7

Biodetection Using Silicon Photonic CrystalMicrocavities

P.M. Fauchet, B.L. Miller, L.A. DeLouise, M.R. Lee, and H. Ouyang

7.1 Photonic Crystals: A Short Introduction

7.1.1 Electromagnetic Theory

Lord Rayleigh was the first to study electromagnetic wave propagation in one-dimensional (1D) periodic media and to identify the angle-dependent, narrowband in which light propagation is prohibited. However, it was not until a fullcentury later, when Yablonovitch [1] and John [2] in 1987 combined Maxwell’sequations with solid-state physics theorems to introduce the concept of pho-tonic bandgaps in two and three dimensions. Many subsequent developmentsin fabrication, theory, and applications (e.g., fiber optics, integrated optics,and negative refraction materials) have since followed.

As shown in Fig. 7.1, photonic crystals (PhCs) are periodically structuredmedia [3] in one, two, or three dimensions. PhCs can be designed to producephotonic bandgaps. Light with photon energies or frequencies that fall insidethis bandgap cannot propagate through the PhC. The periodicity in lengthscale is proportional to the wavelength of light inside the bandgap. PhCs arethe electromagnetic analog of crystalline atomic lattices, in which interferencein the electron wavefunction produces the forbidden bands. Hence, the studyof PhCs is also governed by the Bloch-Floquet theorem.

To study the propagation of light, Maxwell’s equations are solved as aneigenproblem. We assume that light propagates in a mixed dielectric mediumwith no free charges or currents. Maxwell’s equations are then given by:

∇H(r, t) = 0 ∇× E(r, t) +1c

∂H(r, t)∂t

= 0

∇ε(r)E(r, t) = 0 ∇× H(r, t) − ε(r)c

∂E(r, t)∂t

= 0 ,

(7.1)

where ε(r) = εr+ai is the dielectric function, ai is the primitive lattice vectorin all three dimensions, and c is the speed of light in vacuum. By combining

102 P.M. Fauchet et al.

Fig. 7.1. 1D, 2D, and 3D photonic crystals. The periodicity is comparable to thewavelength of light. After Joannopoulos, [3]

the source-free Faraday’s and Ampere’s laws at a fixed frequency ω, one canobtain a harmonic mode solution:

H(r, t) = H(r)eiωt

E(r, t) = E(r)eiωt.(7.2)

The two curl equations can then be written as:

∇× E(r) +iω

cH(r) = 0

∇× H(r) − iω

cε(r)E(r) = 0 ,

(7.3)

leading to the following equation:

∇×(

1ε(r)

∇× H(r))

=(ω

c

)2

H(r). (7.4)

By casting (7.4) as a periodic eigenproblem in analogy with Schrodinger’sequation, the solution can be obtained from the Bloch-Floquet theorem:

H(r) = Hn,k(r)eik·x, (7.5)

with eigenvalues ωn(k) that satisfy:

(∇ + ik) ×(

1ε(r)

(∇ + ik) × Hn,k(r))

=(

ωn(k)c

)2

Hn,k(r). (7.6)

Equation (7.6) yields a different Hermitian eigenproblem over the primi-tive cell at each Bloch wavevector k. These eigenvalues ωn(k) are continuousfunctions of k and can be plotted in a usual dispersion diagram. As the eigen-solutions are periodic functions of k, a numerical computational approach,such as the plane-wave expansion method, is often used to calculate the PhCbandgap diagram in reciprocal space.

7 Biodetection Using Silicon Photonic Crystal Microcavities 103

7.1.2 One-Dimensional and Two-Dimensional PhC

Because Maxwell’s equations are scale-invariant, it is convenient to use nor-malized frequencies ω or wavelengths (a/λ), where a is the lattice constantand λ is the wavelength of light in vacuum. Thus, the same solutions can beapplied to any wavelength simply by choosing the appropriate a. Figure 7.2shows the dispersion curve in a 1D PhC if light propagation is restricted alongthe optical axis z. If the incident wave travels off-axis, the symmetry in thexy plane is destroyed, and the degeneracy of the bands is lifted. This angle ofincidence dependent bandgap imposes practical constraints. Indeed, the takeadvantage of the 1D bandgap, the light beam must be well collimated andaligned along the z-direction.

Let us now consider 2D PhC structures. In these structures, the electro-magnetic field can be decomposed into two polarizations. In the transverseelectric (TE) mode, the E field is in plane, and the H field is perpendicu-lar to the plane. For the transverse magnetic (TM) mode, the situation isreversed. There are two common topologies for making 2D PhCs (Fig. 7.3):square lattice with high index rods surrounded by a low index medium andhexagonal lattice with low-index holes embedded in high index medium. Therod structure is better suited for TM light, and the holes are better suited forTE light. It is also desirable to have bandgaps that overlap in frequency ineach symmetry direction of the crystal. Thus, geometries with a periodicity

Fig. 7.2. Dispersion curve for a 1D PhC, for light propagating on the z-axis only(left) and for light propagating off axis (right). On the right, the thick curves labeledn = 1 correspond to the x-polarized band and the thin curves labeled n = 2 representthe yz-polarized band. When light propagates on axis, the bands are degenerate.After Joannopoulos, Ref. 3


Xa

a

M

Γ

Γ

M K

Fig. 7.3. From left to right : Schematic of 2D PhC topography, Brillouin zone and3D view of a PhC slab. The top row describes a square lattice of dielectric rods inair; the bottom row corresponds to a triangular lattice of air holes in a dielectricsubstrate

Fig. 7.4. Projected band diagram for a finite-thickness slab of air holes in a dielectricslab. The shaded region indicates light cone; only modes below the light cone canpropagate and the rest of the modes suffer from severe radiation loss. The “bandgap”below the light cone exists only for the even (TE-like) modes

that is nearly the same in different directions are preferred, and in the case of2D PhCs, this is an hexagonal lattice.

It is impossible to fabricate a 2D PhC with an infinite slab thickness.An alternative solution is to use thin PhC slabs. The major feature thatdistinguishes the two systems is the light cone, which is a continuum of statesindicated by a shaded region in Fig. 7.4. The modes in the light cone sufferfrom severe radiation loss, while states that lie beneath the light cone areguided modes. These guided modes are confined to the slab because of thehigh refractive index contrast between the slab material (e.g., Si) and thesubstrates (e.g., air or SiO2). The bandgap is then determined by the range


Fig. 7.5. Gap size (as a fraction of the mid-gap frequency) versus Si slab (n = 3.45)thickness for PhC slabs with air hole radius to periodicity ratio r/a of 0.3 and 0.45,respectively. The optimum slab thickness is different for these two cases because ofthe difference in slab effective refractive index

of frequencies for which no guided modes exist. For a slab surrounded byidentical top and bottom layers, two categories of slab modes [even (TE-like)and odd (TM-like)] are then defined according to the reflection symmetrywith respect to the horizontal plane of midthickness of the slab.

The slab thickness plays an important role in determining whether a PhCslab has a bandgap. For example, if the slab is too thin (i.e., less than halfa wavelength), the mode cannot be confined within the slab. As shown inFig. 7.5, the photonic bandgap width depends on the slab thickness, and theoptimal slab thickness varies with different r/a – or different effective slabrefractive index. However, the bandgap width is not be the only considerationin selecting the slab thickness. With the microcavities to be discussed later,longer photon decay times or higher quality factors Q can be achieved byusing slightly thicker slabs.

7.1.3 Microcavities: Breaking the Periodicity

By intentionally introducing a defect in the PhC, localized electromagneticstates can arise inside the photonic bandgap. These localized modes are theoptical analog of the donor or acceptor states produced inside the bandgapof semiconductor crystals. Figure 7.6 illustrates the electric field distributionof a fundamental mode in a PhC microcavity. Instead of staying inside thehigh refractive index Si, the electric field concentrates in lower refractive indexair holes. This property makes it possible for the E-field to interact with theanalyte infiltrated inside the air holes.


Fig. 7.6. Field confinement at the symmetry-breaking defect in a 2D PhC micro-cavity. The electric field is mostly confined in defect region

7.1.4 Computational Algorithms

There are two standard computational approaches for studying PhCs:frequency-domain and time-domain methods. The plane-wave expansion al-gorithm belong to the frequency-domain group. It does a direct computationof the eigenstates and eigenvalues of Maxwell’s equations, and is thus bettersuited for calculating band diagrams. By solving for the eigenvalues, one canprecisely predict eigenstates without missing very closely spaced modes, andeach field is computed at a definite frequency. One of the disadvantages usingthis method is that all of the lowest eigenstates have to be computed beforereaching the states of interest near the bandgap. It is especially problematicin calculating defect states, in which the lower bands are “folded” many times(depending on the array size) in the Brillouin zone. Moreover, frequency-domain approaches require a supercell with periodic boundaries. Hence, alarge array size has to be chosen when computing a single-point-defect PhCmicrocavity because the interactions between repeated supercells must beminimized.

In contrast, the FDTD (finite-domain-time-difference) method, which is atime-domain approach, computes the fields at each computational cycle at adefinite time. This approach is well-suited to compute field distributions andresonance decay-times (quality factor Q). By finding transmission peaks inthe Fourier transform of the system response to the input, resonant modescan be identified. One advantage of the FDTD methods is that by sending apulse into the structure, the response can be obtained at all the frequenciesat the same time. Ideally, FDTD methods can be used to calculate structureswith arbitrary geometries.As the frequency resolution of the Fourier spectrum


is inversely related to the simulation time, a long runtime is required to obtainan acceptable resolution, and the disadvantage in terms of computation timecan be severe.

7.2 One-Dimensional PhC Biosensors

7.2.1 Preparation and Selected Properties of Porous Silicon

One-dimensional PhC biosensors have been made using porous silicon (PSi).PSi is obtained by the electrochemical etching of a crystalline silicon waferin a hydrofluoric acid (HF) solution [4]. The properties of PSi, including thethickness of the porous layer, its porosity, the average pore diameter, andthe pore nanomorphology, can be controlled precisely. As a result, the opticalproperties of PSi can be tuned widely, which makes it a very flexible materialfor optical biosensors. The internal surface of PSi is very large, ranging froma few to hundreds of square meters per gram, which makes PSi very suitablefor capturing biological targets.

The dissolution of silicon requires the presence of fluorine ions (F−) andholes (h+). The pore initiation and growth mechanisms are qualitatively un-derstood. Pore growth can be explained by several models [5–7]. If the sili-con/electrolyte interface becomes rough shortly after etching starts, the sur-face fluctuations of the Si/electrolyte interface either grow (PSi formation) ordisappear (electropolishing). In forward bias for p-type substrates, the holescan still reach the Si/electrolyte interface as the electric field lines are focusedat the tip of the pores. Thus, holes preferentially reach the Si/electrolyte in-terface deep in the pores, where etching can proceed rapidly (Fig. 7.7). Incontrast, no holes reach the end of the Si rods, effectively stopping the etch-ing there. In addition to this electrostatic effect, the random walk of the holestoward the Si/electrolyte interface makes it more likely that they reach it ator near the pore’s tip, also resulting in preferential etching at the pore’s tip.When an n-type substrate is used, porous silicon formation takes place inreverse bias. Another important mechanism becomes predominant if the Sirods are narrow enough (typically much less than 10 nm). In this size regime,the electronic states start to be affected by quantum confinement. When themotion of carriers is restricted in one or more dimensions, the hole states inthe valence band are pushed to lower energy by quantum confinement, whichproduces a potential barrier to hole transport from the wafer to the Si rods.The holes can no longer drift or diffuse into the Si rods and further etchingstops except at the pore’s tip.

When the current density decreases, the number of holes at the pore tipsdrops, which produces smaller pores. Thus, the porosity (defined as the per-centage of void space in the material) can be precisely controlled by the etchingcurrent density. Figure 7.8 shows the dependence of the porosity on current


Fig. 7.7. The dissolution of silicon when holes are injected from the substrate towardthe Si/electrolyte interface. On the left, porous silicon formation takes place whenthe hole flux is relatively small. On the right, electropolishing takes place when thehole flux is large

Fig. 7.8. Dependence of the porosity of the current density for highly-doped n-typesilicon


Fig. 7.9. Top view and cross-sectional SEM images of mesoporous silicon withan average pore diameter of approximately 20 nm, formed in a highly doped p-typesilicon substrate (0.01 ohm-cm), using an electrolyte with 15% HF in ethanol. (c, d):SEM images of 60 nm macropores formed in a very highly doped n-type siliconsubstrate (0.001 ohm-cm) using an electrolyte with 6% HF. (e, f): SEM imagesof 120 nm macropores formed in highly doped n-type silicon substrate (0.01 ohm-cm), using an electrolyte with 6% HF. (g, h): SEM images of 1.5 µm macroporesetched from low doped p-type silicon (20 ohm-cm), using an HF/dimethylformamideelectrolyte

density for highly doped n-type (0.01 ohm-cm) silicon [8]. The pore morphol-ogy is also affected by the choice of doping type and concentration, as illus-trated in Fig. 7.9 [8]. The top view SEM images show PSi samples with differ-ent pore diameters ranging from mesopores (pore size between 10 and 50 nm)to macropores (pore size >50 nm). The bottom-row figures are cross-sectionalSEM images of the same samples. The mesopores formed in p+ silicon sub-strates have very branchy pore walls (Fig. 7.9a, b). The macropores formedin n-type wafers (Fig. 7.9c–h) have much smoother pore walls and larger poresizes. Note that most mesoporous silicon samples and certainly all microp-orous silicon samples (pore sizes <10 nm) exhibit strong luminescence in thevisible to near infrared region [9]. This results from quantum confinement ofelectrons and holes in nanometer-sized quantum structures, which increasesthe bandgap [4,10,11] and enhances the radiative recombination rate [12].

7.2.2 Sensing Principle

PSi is a good host material for label-free optical biosensing applications be-cause its optical properties (photoluminescence and reflectance) are highlysensitive in the presence of chemical and biological species inside the pores [13].PSi optical biosensors with a variety of configurations such as single layerFabry-Perot cavities [14], Bragg mirrors [15], rugate filters [16], and micro-cavities [17–19] have been experimentally demonstrated for the detection oftoxins [20], DNA [17], bacteria [21], and proteins [8, 14, 22]. The capture of


Si Substrate0.6

0

0.2

0.4

0.6

0.8

1

0.7 0.8

Wavelength (µm)PSi Microcavity Wavelength (µm)

b.a. c.

Ref

lect

ance

PL

(a.

u.)

0.9 1 0.7

0

0.2

0.4

0.6

0.8

1

0.8 0.9

Fig. 7.10. (a) A 1D microcavity with a symmetry-breaking defect layer sandwichedbetween two Bragg reflectors; (b) The reflectance spectrum of such a PSi microcavitydisplays a sharp dip near the middle of the high-reflectance bandgap; (c) the pho-toluminescence of bulk PSi consists of a broad band ranging from <700 to �900 m,whereas the PL spectrum of the microcavity consists of a narrow peak centered onthe reflectance dip

biological or chemical molecules inside the pores increases the effective refrac-tive index of the PSi structures, thus resulting in a red shift in the photolu-minescence or reflectance spectra of 1D PSi PhCs.

A PSi microcavity is a 1D photonic band gap structure that contains adefect (symmetry breaking) layer sandwiched between two Bragg mirrors [23].Each Bragg mirror is a periodic stack of layers with two different porosities andquarter wavelength optical thickness. Figure 7.10 shows a PSi microcavity andits typical reflectance and photoluminescence (PL) spectra. Depending on thethickness of the defect layer, the reflectance spectrum contains one or severalsharp resonance dips in the bandgap. The reflectance spectrum modulates themeasured PL spectrum and produces one or several very narrow PL peaks.The position of the reflectance dip or photoluminescence peak is determinedby the optical thickness (refractive index times the physical thickness) of eachlayer in the structure. A slight change of the refractive index inside the porescauses a shift of the spectrum.

The effective dielectric constant of PSi is related to its porosity by theBruggeman effective medium model [24]:

(1 − P )εsi − εPSi

εsi + 2εPSi+ P

εvoid − εPSi

εvoid + 2εPSi= 0, (7.7)

where P is the porosity, εPSi the effective dielectric constant of porous sili-con, εsithe dielectric constant of silicon, and εvoid the dielectric constant ofthe medium inside the pores. Equation (7.7) shows that the effective refrac-tive index of PSi (n2

eff = εPSi) increases as the porosity decreases and asεvoid increases. In Fig. 7.11, neff is plotted as a function of porosity [10]. Insensing applications, εvoid increases due to the binding of targets to the inter-nal surface of the pores. Thus, the overall effective dielectric constant of theporous structure εPSi increases, which causes a red shift in the reflectance dips.


0

1

1.5

2

2.5

Porosity (%)

n-void=1n-void=1.3

n eff

3

3.5

4

20 40 60 80 100

Fig. 7.11. Effective refractive index of porous silicon as a function of porosity fortwo values of the index of refraction of the pores

Figure 7.11 shows that, for a given increase of εvoid, the effective refractiveindex change is larger for higher porosity layers.

To selectively detect the targets of interest, the internal surface of PSineeds to be functionalized. Highly selective elements, such as DNA segmentsand antibodies, can be immobilized on the internal surface of the pores as thebioreceptors or probe molecules. When the sensors are exposed to the target,the probe molecules selectively capture the target molecules. The molecularrecognition events are then converted into optical signals via the increase ofthe refractive index.

7.2.3 One-Dimensional Biosensor Design and Performance

The figure of merit describing the sensitivity of affinity sensors is ∆λ/∆n,where ∆λ is the wavelength shift and ∆n is the change of the ambient re-fractive index. For a PSi microcavity sensor with an average porosity of 75%,∆λ/∆nis ∼ 550 nm [25], which is much larger than for sensing platformsthat rely on the interaction between the evanescent tail of the field and theanalyte [26]. For a system able to detect a wavelength shift of 0.1 nm, theminimum detectable refractive index change of the pores is 2 × 10−4.

The sensitivity of PSi microcavity sensors can be improved by increasingits quality factor Q (Q = λ/∆λ, where λ is the resonance center wavelengthand ∆λ is the full width half maximum of the resonance dip). For higherQ-values, the spectral features are sharper and smaller shifts can be detected.The Q-factor can be increased by increasing the contrast in the porosity of thedifferent layers. However, a higher contrast in porosity is usually produced bya higher contrast in pore opening, which may not be favorable for biosensing


applications when easy infiltration of the target throughout the entire mul-tilayer structure is required. For a given porosity contrast, the Q-factor canalso be increased by increasing the number of periods of the Bragg mirror.In practice, the number of periods cannot be increased arbitrarily for tworeasons. First, uniform infiltration of the molecules becomes more difficult inthicker devices. Second, maintaining a constant HF concentration at the tipof very deep pores is difficult [27], which may lead to an undesired porosityor refractive index gradient.

PSi-based PhC sensors are not perturbed by large, unwanted bioparticles.When the PSi sensor is exposed to a complex biological mixture, only themolecules that are smaller than the pores can be infiltrated into the sensor.Furthermore, an increase of refractive index on the top of the microcavitydue to the presence of large, unwanted objects only causes changes to theside lobes in the reflectivity spectrum, not to the resonance dip [28]. Thus,PSi microcavities are more reliable than planar sensing platforms, where thenonspecific binding of large size objects present in a “dirty” environment mayproduce a false-positive signal.

The pore size also affects the sensitivity of PSi biosensors because thetargets do not completely fill the pores but instead are attached to the porewalls. For a PSi layer of a given porosity, the internal surface area decreasesas the pore size increases. The effective refractive index change of a layer withlarger pores is thus smaller as the percentage of the pore volume occupiedby the biological species is smaller. The spectra for microcavities with fixedporosities (e.g., 80% for the high porosity layer and 70% for the low porositylayer) but different pore diameters (ranging from 20 to 180 nm) have beencalculated. For a given coating thickness (with nlayer = 1.42, a typical valuefor biomolecules), the resonance red shift decreases as the pore size increases,as shown in Fig. 7.12 [8]. For a 0.3 nm thick coating layer, a microcavity with40 nm pores produces a red shift of 10 nm, while a microcavity with 100 nmpores produces a red shift of only 3 nm. Thus, to optimize the sensitivity, thepore size should be as small as possible while still allowing for easy infiltrationof the biological material. Note that the amount of red shift can be used toprecisely measure the amount of material captured inside the pores [29].

7.2.4 Fabrication of One-Dimensional PhC Biosensors

The porosity can be controlled by the etching current density. Once a layerhas been etched, further etching using a different current density does notaffect it, as explained in Sect. 2.1. Stacks of layers with different refractiveindices can thus be formed by switching the current density during etch-ing [30]. Figure 7.13 shows a cross-sectional SEM image of a multilayerstructure etched, using a periodic current density pulse train. The current den-sity determines the porosity and the pulse duration determines the thickness.Figure 7.14 shows top view and cross-sectional SEM images, and reflectancespectra for two different types of microcavities with different pore sizes. The


Fig. 7.12. Calculated red shift of the reflectance dip vs. pore diameter for layers offixed porosities (70 and 80%) and various thicknesses L of the coating on the porewalls

Fig. 7.13. Three low-porosity layers and two high-porosity layers etched in siliconusing a periodic current density. Note the continuity of the pores and the sharpnessof the interfaces between high and low porosity layers. The period is approximately1 µm

mesoporous silicon microcavity with pore diameters from 10 to 50 nm wasfabricated on a p+ wafer (0.01 ohm-cm) with 15% HF in ethanol. The macro-porous silicon microcavity with pore diameters from 80 to 150 nm was etchedin an n+ wafer (0.01 ohm-cm) with 5.5% HF in DI water [10].


a

d e f

b c

200 nm 2 µm

200 nm 2 µm Wavelength (nm)

Wavelength (nm)

Ref

lect

ance

(%

)R

efle

ctan

ce (

%)

Fig. 7.14. Top view and cross-sectional SEM, and reflectance spectra of PSi mi-crocavities made of mesoporous Si (top) and macroporous silicon (bottom)

7.3 Selected Biosensing Results

7.3.1 DNA Detection

Mesoporous silicon microcavities were used for the detection of small DNAsegements (22 base pairs) [17, 31]. Multiple photoluminescence peaks as nar-row as 3 nm of full width half maximum (FWHM) were measured from amicrocavity with a thick defect layer. The first step in the sensor prepara-tion involved the silanization of the thermally oxidized porous silicon samplewith 3-glycidoxypropyltrimethoxy silane, which was hydrolyzed to a reactivesilanol by using DI water (pH∼4). The thermal oxidation treatment not onlyproduced a silica-like surface, but also improved the stability of the lumines-cence. After silanization, the probe DNA, which has an amine group attachedto the 3′ end of the sequence, was immobilized onto the pore surface via dif-fusion. The amine group attacks the epoxy ring of the silane, thereby openingit up and forming a bond. Finally the DNA attached porous silicon sensorwas exposed to the complementary strand of DNA (cDNA). By comparingthe photoluminescence spectrum of the sensor before and after the DNA hy-bridization, the DNA/cDNA binding was detected, as shown in Fig. 7.15 [31].The detection limit of the sensor was determined to be in the picomolar range.

7.3.2 Bacteria Detection

Gram(-) bacteria are responsible for a large numbers of deaths resulting frominfection. To detect Gram(-) bacteria, we selected a target molecule presentin this bacterial subclass and not in Gram(+) bacteria. Lipopolysacharide(LPS) is a primary constituent of the outer cellular membrane of Gram(-)bacteria [32]. LPS is composed of three parts: a variable polysaccharide chain,


600

(a) PSi / DNA

(c) Differential Signal

(b) PSi / DNA/ cDNA

650 700 750

Wavelength (nm)

Nor

mal

ized

PL

Int

ensity

(a.

u.)

800 850 900

Fig. 7.15. Photoluminescence spectra of a mesoporous silicon microcavity with athick defect layer supporting six clearly resolved reflectance dips (PL peaks). (a)After functionalization of the pore surface with probe DNA segments; (b) aftercapture of complementary DNA segments; (c) differential spectrum

a core sugar, and lipid A. As lipid A is common to all LPS subtypes, itis a natural target. An organic receptor, tetratryptophan ter-cyclo pentane(TWTCP) was designed and synthesized as the probe molecule. TWTCPspecifically binds to diphosphoryl lipid A in water with a dissociation constantof 592 nM [59].

After the microcavity sensor was exposed to 3-glycidoxypropyltrimethoxysilane, a mixture of TWTCP and glycine methyl ester was applied to the sen-sor. Glycine methyl ester was used as a blocker molecule to avoid the reactionof the four amino groups of the tetratryptophan receptor with the epoxide-terminated PSi surface. As shown in Fig. 7.16, upon exposure of lysed Gram(-)cells (Escherichia coli) to the immobilized TWTCP sensor, a 4 nm red shift ofthe photoluminescence spectrum of the microcavity was detected [21]. How-ever, when the sensor was exposed to a solution of lysed Gram(+) cells (Bacil-lus subtilis), no shift of the spectrum was observed. These results were con-firmed with all Gram(-) and Gram(+) bacteria tested.

7.3.3 Protein Detection

A protein biosensor based on macroporous silicon microcavity was demon-strated with the streptavidin-biotin couple [8]. Biotin is a small molecule, whilestreptavidin is relatively large (67 kDa), making its infiltration difficult into


1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

0.6

0.4

0.2

0

−0.2

−0.4

−0.6

675 775

Wavelength (nm)

Inte

nsity

875

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0675 775

Wavelength (nm)

Inte

nsity

Inte

nsity

0.6

0.4

0.2

0

−0.2

−0.4

−0.6

Inte

nsity

875

675 775

Wavelength (nm)

875 675 775

Wavelength (nm)

875

Fig. 7.16. Photoluminescence spectra of a mesoporous silicon biosensor similar tothe one shown in Fig. 7.15 before and after exposure to bacterial cell lysates (top row)and differential spectra (bottom row). The spectra on the left have been obtainedfor bacteria that are Gram(+) and the spectra on the right for bacteria that areGram(-)

mesopores but easy into macropores. Furthermore, each streptavidin tetramerhas four equivalent sites for biotin (two on each side of the complex), whichmakes it a useful molecular linker.

To create a biotin-functionalized sensor for the capture of streptavidin,microcavities were first thermally oxidized, then silanized with aminopropy-ltriethoxysilane to create amino groups on the internal surface. The probemolecules, Sulfo-NHS-LC-LC-biotin, were then immobilized inside the pores.As shown in Fig. 7.17a, the red shift of the resonance increased with increas-ing biotin concentration. A 10 nm red shift corresponds to a nearly complete(95%) biotin surface coverage.


14 20

16

12

8

4

0

100

80

60

40

20

0

12

10

8

6

4

2

00 0.4 0.8

Biotin concentration (mg/ml) Wavelength (nm) Biotin surface coverage (%)

∆λ (nm

)

∆λ (nm

)

Ref

lect

ivity

(%)

1.2 1.6 2 600 650 700 750 800 850 900 0 20 40 60 80 100

(a) (b) (c)

Fig. 7.17. (a) Increase of the reflectance dip red shift when more probe molecules(biotin) are added; the shift eventually saturates when all the pore walls are coveredby biotin. (b) After exposure to target molecules (streptavidin), the microcavityspectrum shifts to the red. (c) The measured red shift upon capture of the targetsdepends on the coverage of the pore walls by the probe molecules

To study how the biotin surface coverage affects the binding to strepta-vidin, sensors derivatized with different biotin concentrations were exposed tothe same amount of the target: 50 ml of streptavidin with a concentration of1 mg ml−1. After exposure to streptavidin, a red shift was detected, which is at-tributed to the specific binding of streptavidin to the biotin-derivatized macro-porous microcavity (Fig. 7.17b). For the control samples that were silanizedbut did not contain biotin, no shift was detected after exposure to strepta-vidin, which indicates that nonspecific absorption of streptavidin inside themicrocavity is nonexistent or negligible.

The red shift caused by streptavidin binding to biotin is a function of thebiotin surface coverage. Figure 7.17c shows that there is an optimum biotinsurface coverage that maximizes the red shift, and therefore the capture ofstreptavidin. This behavior is in good agreement with observations by otherresearchers using different sensing techniques [34]. The spacing between eachneighboring biotin needs to be large enough so that biotin can reach thepocket-like binding site of streptavidin. The roll-off of the red shift showingin Fig. 7.17c is due to the fact that when the biotin surface density becomesvery high and each biotin is closely surrounded by its neighbors, biotin cannotprotrude deep enough into the binding site of streptavidin, thus decreasing theprobability of capturing streptavidin. Other proteins have also been detectedusing 1D PhC microcavities made of PSi, including intimin, a protein secretedby pathogenic E. coli [22].

7.3.4 IgG Detection

Immunoglobulin (IgG) is the most common type of antibody synthesized inresponse to a foreign substance (antigen). Antibodies have a specific molecularstructure capable of recognizing a complementary molecular structure on theantigen, such as proteins, polysaccharides, and nucleic acids. From the X-raycrystal structure, the longest dimension of IgG is approximately 17 nm [35].


Fig. 7.18. A PSi microcavity biosensor functionalized for the capture of one typeof IgG can discriminate against another type of IgG with a contrast ratio �10×

The detection of rabbit IgG (150 kDa) was investigated through multiple lay-ers of biomolecular interactions in a macroporous silicon microcavity sensor.The silanized sensor was first derivatized with biotin, which can selectivelycapture streptavidin. Exposure of biotinylated goat anti-rabbit IgG to thesensor results in its attachment to the surface through the binding betweenbiotin and streptavidin. The sensor used goat anti-rabbit IgG as the probemolecule to selectively capture rabbit IgG. A red shift of the spectrum canbe detected when each layer of molecules is added to the sensor. As shown inFig. 7.18 when the sensor was exposed to 50 ml of a solution containing rabbitIgG at a 1 mg ml−1 concentration, a 6 nm red shift was detected [25]. Whenthe sensor was exposed to 50 ml goat IgG (1 mg ml−1), which does not bindto the goat anti-rabbit IgG, the red shift was negligible (<0.5 nm).

7.4 Two-Dimensional PhC Biosensors

7.4.1 Sample Preparation and Measurement

Silicon-on-insulator (SOI) wafers were used to fabricate 2D PhC microcavitybiosensors [36]. The Si slab thickness was approximately 400 nm. Rigorous 3Dsimulations using the FDTD method were performed prior to the fabrication.The incident light was coupled along the Γ -M direction because the in-planeleakage of the resonance mode is mainly in the Γ -M direction and hence thecoupling efficiency is higher along the Γ -K direction. The ridge waveguidewas tapered down from 2 to 0.6 µm with an external surface cross-sectionof 2 µm× 400 nm, which predominantly select fundamental TE mode as thepropagation mode.


Fig. 7.19. Fabrication flow starting from a silicon on insulator (SOI) wafer andending with the 2D PhC microcavity inserted in a silicon wire waveguide

Fig. 7.20. Experimental setup for measuring the transmission of a 2D PhC micro-cavity. A microscope objective is used to observe the alignment of all the components

The fabrication procedure is depicted in Fig. 7.19. After the SOI waferwas cleaned and an oxide mask grown, negative (PMMA) and positive (FOX)resists were used for pattern transformation. After electron-beam lithographyand resist development, reactive ion etching was performed to transfer thepattern onto the Si slab. The ridge waveguide facets were then polished toimprove the coupling efficiency.

Figure 7.20 illustrates the measurement setup. An HP8168F laser, tunablefrom 1,440 to 1,590 nm, was used as the light source. The TE polarized laserbeam was focused onto the input tapered ridge waveguide (from 2 µm to0.6 µm) using a tapered lensed fiber (with a spot size of 2 µm at focus), and thetransmitted signal was measured by a photodetector. To optimize the intensityof the TE polarized-light, a polarization controller was inserted between thelaser and the input fiber.

7.4.2 Sensing Principle

The sensing principle is very similar to the one used in the 1D PhC biosen-sors. In contrast to chemical sensors [37], in biological sensors, the biomolecule


Fig. 7.21. After functionalization of the air hole surfaces with probe molecules(left), targets are captured by the biosensor (middle), which produces a coating onthe air hole walls (right)

recognition strongly depends on the surface chemistry, thus, instead of fillingup the pores, the molecules prefer to form a monolayer coating at the pore(air hole) wall As shown in Fig. 7.21, highly selective probes (e.g., DNA, an-tibody) can be immobilized on the internal surface of the air holes and forma monolayer that can capture specific target molecules (e.g., matching DNA,proteins). Hence, the coating causes a refractive index change only in thevicinity of the pore wall.

7.4.3 Selected Biosensing Results

Protein Detection

To test the device performance, glutaraldehyde-bovine serum albumin (BSA)coupling was used as the model system. The air hole is ∼30 times larger thanthe BSA hydrodynamic diameter [38], so that the infiltration proceeds easily,and both glutaraldehyde and BSA can form a uniform monolayer coated onthe pore walls.

To prepare the surface for capturing biomolecules, the device was first ther-mally oxidized at 800◦C to form a silica-like interface for binding of aminogroups. To functionalize the sensor and capture glutaraldehyde, the microcav-ity internal surface was silanized. Subsequently, the target, BSA, was immo-bilized covalently on the sidewall. The experiment protocol is as follows: (1)clean the sensor surface with DI water and dry it under nitrogen flow. Storethe sensor in moist ambient; (2) drop 5 µl of 2% aminopropyltrimethoxysilaneon the sensor for 20 min; (3) rinse the sensor with DI water and dry withnitrogen flow, and then bake at 100◦C for 10 min; (4) drop 5 µl of 2.5% glu-taraldehyde in Hepes buffer on the sensor for 30 min; soak in buffer for 10 minto dilute the unreacted agents and dry it under nitrogen flow; (5) drop 5 µl of1% BSA on the sensor for 30 min; then soak in buffer for 20 min and dry itwith nitrogen.


Fig. 7.22. Transmission near the 2D PhC microcavity resonance for the function-alized sensor (a), after exposure to the probe, glutaraldehyde (b), and subsequentexposure to the target, bovine serum albumin

Figure 7.22 illustrates the normalized transmission spectra measured atthree different binding stages. The PhC microcavity consisted of a triangulararray of cylindrical air pores in a 400 nm-thick silicon (Si) slab. The latticeconstant a was 465 nm, and the pore diameter r was 0.3a. A defect was intro-duced by reducing the center pore diameter to 0.18a, leading to a transmis-sion resonance in the bandgap close to 1.58 µm. Curve (a) shows the initialtransmission after the oxidation and silanization. Curve (b) was measured af-ter exposure to glutaraldehyde. A resonance red shift of 1.1 nm is observed.Curve (c) shows a red shift of 1.7 nm after BSA binding corresponding to atotal shift of 2.8 nm with respect to the initial spectrum. To verify the exper-imental results, a plane-wave expansion calculation with 32 grid points persupercell was performed. The simulation of the resonance red shift assumesthat the refractive index of the dehydrated proteins is 1.45. This value agreeswith an ellipsometry measurement [29] and with the literature values [39].

Figure 7.23 plots the predicted resonance red shift vs. coating thickness.Using ellipsometry, the thickness of the dehydrated glutaraldehyde monolayerwas measured as 7± 1 A. We then calculated that glutaraldehyde should intro-duce a resonance red shift of 0.98± 0.2 nm, which is in good agreement withthe experimental data. The thickness of a dehydrated BSA layer measuredwith the same method is approximately 15± 5 A, which should introducean additional red shift of 2.7± 1 nm or a total red shift of 3.8± 1 nm. Theexperimental data are again in good agreement with the model.


0 10 200

2

4

6

8

30

Coating thickness t (Å)

Red

shift

∆λ

(nm

)

40 50

Fig. 7.23. Calculated red shift vs. coating thickness of the air hole walls for thedevice structure tested in Fig. 7.22

The biomolecules can also attach to the sidewall, the bottom, or the topof the device. 3D simulations (Fig. 7.24) using FDTD methods were carried tosimulate the molecule-position dependence. Within the molecule size range ofinterests (i.e. <50 nm), the coating on the pore wall contributes the most tothe spectral shift, whereas the presence of biomolecules elsewhere produces amuch smaller shift. This is not surprising, as the mode is strongly confined inthe Si slab and on the air hole sidewalls, the interaction of the mode and theanalyte inside the device is much stronger than on top of the device.

Toward Single Molecule Detection

As the electric field is strongly localized in the defect region, what wouldhappen if the biomolecules were present only in the defect region? [40] Thesensitivity ∆λ/∆t decreases by a factor of 4, but the total amount of ana-lyte that is detectable drops by nearly two orders of magnitude, from 2.5 pgto 0.05 pg. To demonstrate the potential for single bioparticle detection, amicrocavity with a large defect (diameter ∼900 nm) was fabricated. Latexspheres (density of 1.05 g cm−3, refractive index n = 1.59) with a diameterof approximately 370 nm were dropped on the PhC. As shown in Fig. 7.25,


100

80

60

40

20

00 10 20 30 40

Capable of detecting amount of analyte ~ 2.5 fg

Si

Coating thickness ∆t (nm)

Res

onan

ce s

hift

∆λ (

nm)

50 60 70

Fig. 7.24. Calculated resonance red shift vs coating thickness, for a biomoleculelayer on the side walls of the air holes, at the bottom of the air holes, or on top ofthe device

Fig. 7.25. Close up of near the central defect of a 2D PhC microcavity in whichone latex microsphere has been captured

one latex microsphere was inside the defect region and the rest of the spheresremained on top of the device. The normalized transmission spectra of thePhC microcavity shown in Fig. 7.26 demonstrate that the device can detect asingle bioparticle. For a device with a quality factor Q of ∼2,000, a spectralshift of 1 nm can be measured very easily. Simulations show that biomoleculessmaller than 50 nm can be detected.


Fig. 7.26. Transmission near the resonance of the 2D PhC microcavity when thedefect air hole contains one latex microsphere (curve b) or no microsphere (curvea). The microsphere’s diameter was 370 nm

7.5 Conclusions

PhC microcavities made of silicon are a platform of choice for biosensing.After proper functionalization, both 1D and 2D devices have excellent per-formance in terms of sensitivity and selectivity. It is important to note thata complete biosensor system requires other components (such as prefiltrationand concentration modules [41]) that have not been described here but canalso be made in silicon and integrated with the PhC microcavity.

Acknowledgments

The work described in this chapter was supported by the US National Sci-ence Foundation through grant BES 04279191. Fabrication of the 2D PhCmicrocavities was performed in part at the Cornell NanoScale Facility, whichis supported by NSF through grant ECS 03-35765. The technical contribu-tions of Dr. Selena Chan, Dr. Marc Christophersen, and Dr. Chris Striemerare gratefully acknowledged.

References

1. E. Yablonovitch, Phys. Rev. Lett., 58, 2059 (1987)


2. S. John, Phys. Rev. Lett., 58, 2486 (1987)3. J.D. Joannopoulos, R.D. Meade, J.N. Winn, in Photonic Crystals: Molding the

Flow of Light, (Princeton Press, 1995)4. P. M. Fauchet, in Encyclopedia of Applied Physics, Update 2, (Wiley-VCH

Verlag, 1999), pp. 249–2725. R.L. Smith, S.D. Collins, J. Appl. Phys. 71, R1 (1992)6. A.G. Cullis, L.T. Canham, P.D.J. Calcott, J. Appl. Phys. 82, 909 (1997)7. V. Lehmann, Electrochemistry of Silicon, (Wiley-VCH, 2002)8. H. Ouyang, M. Christophersen, R. Viard, B.L. Miller, P.M. Fauchet, Adv. Funct.

Mater. 15, 1851 (2005)9. P.M. Fauchet, in Semiconductors and Semimetals, vol. 49, ed. by D.J. Lockwood,

(Academic Press, San Diego, 1998), pp. 206–25210. L.T. Canham, Appl. Phys. Lett. 57, 1046 (1990)11. V. Lehmann, U. Gosele, Appl. Phys. Lett. 58, 856 (1991)12. M.S. Hybertsen, Phys. Rev. Lett. 72, 1514 (1994)13. M.P. Stewart, J.M. Buriak, Adv. Mater. 12, 859 (2000)14. K.-P.S. Dancil, D.P. Greiner, M.J. Sailor, J. Am. Chem. Soc. 121, 7925 (1999)15. P.A. Snow, E.K. Squire, P.S.J. Russell, L.T. Canham, J. Appl. Phys. 86, 1781

(1999)16. F. Cunin, T.A. Schmedake, J.R. Link, Y.Y. Li, J. Koh, S.N. Bhatia, M.J. Sailor,

Nat. Mat. 1, 39 (2002)17. S. Chan, Y. Li, L.J. Rothberg, B.L. Miller, P.M. Fauchet, Mat. Sci. Eng. C 15,

277 (2001)18. L. Pavesi, La Rivista del Nuovo Cimento 20, 1 (1997)19. H. Ouyang, M. Christophersen, P.M. Fauchet, Phys. Stat. Sol. (a) 202, 1396

(2005)20. H. Sohn, S. Letant, M.J. Sailor, W.C. Trogler, J. Am. Chem. Soc. 122, 5399

(2000)21. S. Chan, S.R. Horner, P.M. Fauchet, B.L. Miller, J. Am. Chem. Soc. 123, 11797

(2001)22. H. Ouyang, L.A. DeLouise, B.L. Miller, P.M. Fauchet, Anal. Chem. 79, 1502

(2007)23. J.E. Lugo, H.A. Lopez, S. Chan, P.M. Fauchet, J. Appl. Phys. 91, 4966 (2002)24. W. Theiβ, Surf. Sci. Rep. 29, 91 (1997)25. H. Ouyang, M. Lee, B.L. Miller, P.M. Fauchet, Proc. SPIE 5926, 59260J (2005)26. J. Scheuer, W.M.G. Green, G.A. DeRose, A. Yariv, IEEE J. Sel. Top. Quant.

Electron. 11, 476 (2005)27. M. Thonissen, M.G. Berger, S. Billat, R. Ares-Fischer, M. Kruger, H. Luth,

W. Theiss, S. Hillbrich, P. Grosse, G. Lerondel, U. Frotscher, Thin Solid Films297, 92 (1997)

28. H. Ouyang, L.A. DeLouise, M. Christophersen, B.L. Miller, P.M. Fauchet, Proc.SPIE 5511, 71 (2004)

29. H. Ouyang, C.C. Striemer, P.M. Fauchet, Appl. Phys. Lett. 88, 163108 (006)30. S. Chan, P.M. Fauchet, Appl. Phys. Lett. 75, 274 (1999)31. S. Chan, P.M. Fauchet, Y. Li, L.J. Rothberg, B.L. Miller, Phys. Stat. Sol. (a)

182, 541 (2000)32. L.S. Young, W.J. Martin, R.D. Meyer, R.J. Weinstein, E.T. Anderson, Ann.

Inern. Med. 86, 456 (1997)33. R.D. Hubbard, S.R. Horner, B.L. Miller, J. Am. Chem. Soc. 123, 5811 (2001)


34. L.S. Jung, K.E. Nelson, P.S. Stayton, C.T. Campbell, Langmuir 16, 94221(2000)

35. E.O. Saphire, P.W.H.I. Parren, R. Pantophlet, M.B. Zwick, G.M. Morris, P.M.Rudd, R.A. Dwek, R.L. Stanfield, D.R. Burton, I.A. Wilson, Science 293, 1155(2001)

36. M. Lee, P.M. Fauchet, Opt. Exp. 15, 4530 (2007)37. E. Chow, A. Grot, L.W. Mirkarimi, M. Sigalas, G. Girolami, Opt. Lett. 29, 1093

(2004).38. I.D. Kuntz, W. Kauzmann, Adv. Protein Chem., 28, 239 (1974)39. M. Malmsten, J. Colloid. Interf. Sci., 166, 333 (1994)40. M.R. Lee, P.M. Fauchet, Paper presented at the 4th IEEE LEOS International

Conference on Group IV Photonics, Tokyo, Japan, September 200741. C.C. Striemer, T.R. Gaborksi, J.L. McGrath, P.M. Fauchet, Nature 445, 749

(2007)

8

Optical Coherence Tomographywith Applications in Cancer Imaging

S.A. Boppart

8.1 Introduction

Optical coherence tomography (OCT) is a rapidly emerging optical imag-ing technique for a wide range of biological, medical, and material investiga-tions [1,2]. OCT was initially developed in the early 1990s, and has providedresearchers with a novel means by which biological specimens and nonbiolog-ical samples can be visualized. A primary advantage of OCT is the ability toimage tissue microstructure in situ at micron-scale image resolution, withoutthe need for excision of a specimen for tissue processing. The optical rangingin OCT is analogous to ultrasound B-mode imaging, except that OCT useslow-coherence light rather than high-frequency sound. Cross-sectional OCTimages can be generated, as is commonly done in ultrasound, or en face OCTsections can be acquired, as in confocal and multiphoton microscopy. TheOCT imaging principle involves optical ranging, where the optical reflectionof light from a low-coherence optical source is spatially localized using in-terferometry. The OCT image that is assembled is a gray-scale or false-colormultidimensional spatial representation of backscattered light intensity. Thesignal intensity represented within an OCT image represents the differentialbackscattering contrast between different tissue types on a micron scale. OCTperforms imaging using light; therefore, it has a one to two order-of-magnitudehigher spatial resolution than ultrasound. Because the optical imaging beamcan be transmitted readily through air, OCT beam-delivery systems do notrequire contact with the specimen or sample, as do ultrasound probes. Spectro-scopic characterization of tissue and cellular structures is also possible withinthe optical spectrum of the light source.

Ophthalmology was the first medical application area for OCT, wherehigh-resolution tomographic imaging of the anterior eye and retina is possi-ble [3]. The transparency of the eye at visible and near-infrared wavelengths,and its accessibility using optical instruments and techniques, has enabledextensive research and clinical applications for diagnostic OCT imaging. The

128 S.A. Boppart

diagnosis of many retinal diseases is possible because OCT can provide imagesof retinal pathology with micron-scale resolution.

OCT has also been applied in a wide range of nontransparent tissues [4,5].In nontransparent tissues such as skin, muscle, and other soft tissues, theimaging depth is limited by optical attenuation due to scattering and ab-sorption. Optical scattering decreases with increasing wavelength. Therefore,while ophthalmic OCT imaging has primarily been performed at 800 nmwavelengths, OCT imaging in nontransparent tissues has been typically per-formed with wavelengths of 1.0–1.3 µm. Imaging depths up to 2–3 mm canbe achieved using a system detection sensitivity of 100–110 dB. In early ex-ploration imaging studies, OCT has been performed in virtually every organsystem to investigate applications in cardiology [6–8], gastroenterology [9,10],urology [11, 12], neurosurgery [13], and dentistry [14], to name a few. Us-ing short coherence length, short-pulsed light sources, high-resolution OCThas been demonstrated with axial resolutions of less than 5 µm [15–17]. High-speed real-time image acquisition rates have also been achieved, which enablesvolumetric imaging [18, 19]. New imaging modes in OCT have been demon-strated, such as Doppler OCT imaging of blood flow [20–22] and birefringenceimaging to investigate laser intervention [23–25]. OCT beam delivery systemsincluding transverse imaging catheter/endoscopes and forward imaging de-vices have enabled OCT imaging of internal structures [26–28], and mostrecently, catheter-based OCT has been used to perform in vivo imaging inanimal models and human patients [29–33].

8.2 Principles of Operation

OCT performs optical ranging in tissue, using high spatial resolution andhigh dynamic range detection of backscattered light as a function of opticaldelay. While the speed of ultrasound waves in tissue is relatively slow anddetectable using electronics, the velocity of light is extremely high. There-fore, the time delays of the reflected light cannot be measured directly andinterferometric detection techniques must be employed. Low-coherence inter-ferometry or optical coherence domain reflectometry are techniques that areused in OCT. First developed in the telecommunications field for measuringoptical reflections from faults or splices in optical fibers [34], low-coherence in-terferometry can also be used for localizing the optical reflections in biologicaltissue. Subsequently, the first applications in biological samples included one-dimensional optical ranging in the eye to determine the location of differentocular structures [35,36].

Time delays of reflected light-off of tissue boundaries are typically mea-sured using a Michelson-type interferometer (Fig. 8.1). Other interferometerdesigns, such as a Mach–Zehnder interferometer, have been implemented tooptimize the delivery of the OCT beam and the collection of the reflectedsignals [37,38]. Light reflected from the specimen or sample is interfered with

8 OCT with Applications in Cancer Imaging 129

Fig. 8.1. Schematic illustrating the concept of low coherence interferometry. Using ashort coherence length light source and a Michelson-type interferometer, interferencefringes are observed only when the path lengths of the two interferometer arms arematched to within the coherence length of the light source in a time-domain OCTsystem. The full-width half-maximum of the envelope of the autocorrelation functionis equal to the coherence length (∆lc) and axial resolution in OCT. This envelopealso represents the axial point-spread-function of an OCT system

light reflected from a reference path of known path length, which spatiallydetermines the location of the reflection in depth. Interference of the lightbetween the two arms of the interferometer can occur only when the opticalpath lengths of the two arms match within the coherence length (axial resolu-tion) of the optical source. If the reference arm optical path length is scanned,as in a time-domain OCT system, different delays of backscattered light fromwithin the sample are measured, and a single column of depth-dependent datais collected. The interference signal is detected at the output port of the in-terferometer, digitized, and stored on a computer. Following a depth (z) scan,the incident beam is scanned in the transverse direction (x) and multiple ax-ial measurements are performed. A two-dimensional data array is generated,which represents the optical backscattering through a cross-sectional plane inthe specimen (Fig. 8.2). Similarly, the OCT beam can be translated in thethird (y) dimension, and a series of two-dimensional cross-sectional imagescan be collected to form a three-dimensional volume. The logarithm of thebackscatter intensity is then mapped to a false-color or gray-scale and dis-played as an OCT image. Typically, the interferometer in an OCT instrumentcan be implemented using fiber optic couplers, and beam-scanning can be

130 S.A. Boppart

Fig. 8.2. An OCT image is based on the spatial localization of variations in op-tical backscatter from within a specimen. For depth-priority scanning, images areacquired by performing axial measurements of optical backscatter at different trans-verse positions on the specimen and displaying the resulting two-dimensional dataset as a gray-scale or false-color image. OCT images can also be acquired withtransverse-priority scanning to collect an en face image similar to optical section-ing in confocal or multiphoton microscopy. Multiple planes in either mode can beassembled for three-dimensional OCT

performed with small mechanical galvanometers or another beam-delivery in-strument to yield a compact and robust system (Fig. 8.3). The axial resolutionin OCT images is determined by the coherence length of the light source, incontrast to the tight focus and confocal region in confocal microscopy. Theaxial point spread function of the OCT measurement is defined by the sig-nal detected at the output of the interferometer (detector arm), and is theelectric-field autocorrelation of the source. The coherence length of the lightis the spatial width of the field autocorrelation, and the envelope of the fieldautocorrelation is equivalent to the inverse Fourier transform of the opticalsource power spectrum. The width of the autocorrelation function (axial res-olution), therefore, is inversely proportional to the width of the optical powerspectrum. If the optical source has a Gaussian spectral distribution, then thefree-space axial resolution ∆z is given by

∆z =2 ln 2

π

λ2

∆λ, (8.1)

where ∆z and ∆λ are the full widths at half maximum of the autocorrelationfunction and power spectrum, respectively, and λ is the central wavelength of


Fig. 8.3. Schematic representation of an OCT system implemented using fiber op-tics. The Michelson interferometer is implemented using a fiber-optic coupler. Lightfrom the low-coherence source is split and sent to a sample arm with a beam deliv-ery instrument and a reference arm with an optical path-length scanner. Reflectionsfrom the arms are combined and the output of the interferometer is detected witheither a photodiode or a linear CCD array in a spectrometer. Components for botha time-domain and a spectral-domain OCT system are shown

the source. The dependence of the axial resolution on the bandwidth of theoptical source is plotted in Fig. 8.4. To achieve high axial resolution (approach-ing 1 µm), therefore, requires extremely broad bandwidth optical sources. Thecurves plotted in Fig. 8.4 are for three commonly used wavelengths, 800, 1,300,and 1,500 nm. From (8.1), higher resolutions can be achieved with shorterwavelengths. However, shorter wavelengths are more highly scattered in bio-logical tissue, and frequently result in less imaging penetration.

Evident from this discussion, the axial and transverse resolutions in OCTare not directly related, as in conventional microscopy. However, the transverseresolution in an OCT imaging system is determined by the focused spot size,according to the principles of Gaussian optics, and is given by

∆x =4λ

π

f

d, (8.2)

where d is the beam diameter on the objective lens and f is the focal lengthof the objective lens. Large numerical aperture optics can be used to fo-cus the beam to a small spot size and provide high transverse resolution.

132 S.A. Boppart

00

5

10

15

20

25

30

50 100 150 200Source Spectral Bandwidth (nm)

1500 nm1300 nm800 nm

Co

her

ence

Len

gth

(mm

)

250 300 350

Fig. 8.4. Dependence of coherence length (axial resolution) on optical source band-width. Curves are plotted for 800, 1,300, and 1,500 nm, three common wavelengthsused in OCT. High axial imaging resolution is achieved with broad spectral band-widths and shorter wavelengths. Shorter wavelengths, however, are more highly ab-sorbed in biological tissue, decreasing imaging penetration depth

The transverse resolution is also related to the depth of focus or the confocalparameter 2zR (two times the Raleigh range).

2zR =π∆x2

2λ. (8.3)

Increasing the transverse resolution (decreasing the spot size at the focus)subsequently results in a reduced depth of field. For OCT imaging, the con-focal parameter or depth of focus is typically chosen to match the desireddepth of imaging. High transverse resolutions are often required and may beutilized in OCT. However, the short depth of field requires additional opticalor mechanical techniques to spatially track the focus in depth along with theaxial OCT scanning.

Relatively small incident powers of 1–10 mW are commonly required forOCT imaging. Typically, the real-time (∼30 frames per second) acquisitioncan be achieved at a signal-to-noise ratio of ∼100 dB with 5–10 mW of incidentoptical power. Recent advances in spectral-domain OCT (SD-OCT) [39, 40]and swept-source OCT (SS-OCT) [41, 42] have enabled extremely fast axialscanning and real-time volumetric imaging at micron-scale resolution, whilemaintaining sufficiently high SNR for imaging larger volumes of tissue.


8.3 Optical Sources for Optical Coherence Tomography

OCT imaging systems have primarily used superluminescent diodes (SLDs)as low coherence light sources [43]; however, new sources are becoming widelyavailable. SLDs are attractive because they are compact, have high efficiency,low noise, and are commercially available at a range of wavelengths including800 nm, 1.3 µm, and 1.5 µm. Output powers, however, are typically less than afew milliwatts, which frequently limits the use of these sources to slow acquisi-tion rates in order to preserve signal-to-noise ratio. The available bandwidthsare relatively narrow, permitting imaging with 10–15 µm resolution. Becauseof the need for broader bandwidth sources, manufacturers have been devel-oping SLD sources that have bandwidths of 50–100 nm (3–10 µm resolution,depending on wavelength).

Advances in short-pulse, solid-state, laser technology make these sourcesattractive for OCT imaging in research applications. Femtosecond solid-statelasers can generate tunable, low-coherence light at powers sufficient to per-mit high-speed OCT imaging [15–17]. The titanium:sapphire (Ti:Al2O3) laserfrom 0.7 to 1.1 µm has been a commonly used source, not only for OCT, butmost often for multiphoton microscopy. Imaging resolutions of less than 1 µmhave been demonstrated at 800 nm using a Ti:Al2O3 laser source [17]. Todemonstrate the improvement in resolution afforded by a titanium:sapphirelaser, a comparison between the power spectra and autocorrelation func-tions of a 800 nm center wavelength SLD and a short pulse (≈5.5 fs) tita-nium:sapphire laser is shown in Fig. 8.5. An order-of-magnitude improvementin axial resolution is noted for the short pulse titanium:sapphire laser source[16]. In an effort to develop more compact and convenient sources comparedto a titanium:sapphire laser, superluminescent fiber sources have been inves-tigated [44, 45] as well as pumped optical fibers that can produce extremelybroad supercontinuum [46–48]. Since the titanium:sapphire (Ti:Al2O3) lasertechnology is routinely used in multiphoton microscopy applications for itshigh peak intensities to enable multiphoton absorption and subsequent emis-sion of fluorescence from exogenous fluorescent contrast agents [49], the si-multaneous generation of OCT and multiphoton microscopy images has beendemonstrated, providing complementary image data using a single opticalsource [50].

8.4 Fourier-Domain Optical Coherence Tomography

OCT has most-commonly been performed by rapidly scanning the referencearm path-length to acquire a single scan in depth. This mode has been termedtime-domain OCT (TD-OCT). In this mode, faster image acquisition rateswere therefore dependent on the rate at which this optical path-length couldbe varied, and rapid-scanning optical delay lines were previously developed for

134 S.A. Boppart

Fig. 8.5. Comparison of optical output spectra (top) with interference signals(lower left) and envelopes (lower right) for a Kerr-lens modelocked titanium:sapphire(Ti:Al2O3) laser vs. a superluminescent diode (SLD). The broad optical bandwidthof the titanium:sapphire laser (260 nm) permits a free-space axial resolution of1.5 µm. In comparison, the superluminescent diode with a 32 nm spectral bandwidthpermits an axial resolution of 11.5 µm. Figure reprinted with permission [16]

this purpose. In the mid 1990s, it was realized, and in recent years, demon-strated that alternatives to this time-domain mode of scanning were tech-niques that simultaneously collected light from all depths into the tissue asthe beam was positioned at each transverse position. The light sources forthese techniques are either a broadband source in which the collected light isdetected by a linear CCD array in a spectrometer (spectral-domain OCT, SD-OCT) [39,40], or a narrow-band source which is rapidly swept over a range offrequencies and the collected light is detected by a photodiode (swept-sourceOCT, SS-OCT) [41, 42]. For both of these techniques, depth-dependent scat-tering information is obtained by taking the inverse Fourier transform of thisacquired spectral data. Recently, these techniques have become preferred forthe improvement in acquisition rate and sensitivity that they provide over thetime-domain OCT systems.

Recent use of the SS-OCT has demonstrated real-time acquisition rates.The primary advantages to using a swept-source is that all the optical poweris contained within a narrow wavelength range at any one point in time, and


then is swept to a different wavelength. Wide tuning ranges make high imagingresolution possible and at the rate of hundreds of thousands of axial scans persecond, fast volumetric imaging can be performed. The development of fastswept sources has utilized innovative methods, including a rotating polygonalmirror array [41] or Fourier-domain mode-locking [42].

High-speed SD-OCT has been demonstrated in the living human eye andin other real-time applications. Because the acquisition rate is no longer de-pendent on a mechanically scanned reference arm, significantly faster rates areachievable, up to typical rates of 30,000 axial scans per second, depending onthe read-out rates of the linear CCD arrays in the spectrometer. The computa-tional complexity has been increased for both SD-OCT and SS-OCT becausethe inverse Fourier transform of the acquired data is required for each axialscan line. Because dedicated digital-signal-processing (DSP) chips are readilyavailable, it appears that the computational power necessary for this appli-cation is feasible and available to the research investigator. Spectral-domainOCT also has the advantages of improved phase stability, because no mechan-ical scanning is performed, and higher signal-to-noise ratio, because individualwavelengths are now detected by separate elements in a linear detector array.Linear detector arrays, especially indium–gallium–arsenide arrays used for de-tecting the commonly used 1,300 nm OCT wavelength, are expensive, shiftingthe costs of rapid scanning optical delay lines for time-domain systems to thedetector array for SD-OCT.

8.5 Beam Delivery Instruments for Optical CoherenceTomography

The OCT imaging technology is modular in design. This is most evidentin the optical instruments through which the OCT beam can be deliveredto the tissue. Because OCT is fiber-optic based, single optical fibers can beused to deliver the OCT beam and collect the reflected light, thereby mak-ing the beam delivery system potentially very small, on the order of thesize of an optical fiber (125 µm diameter) itself. The OCT technology canalso be readily integrated into existing optical instruments such as researchand surgical microscopes [51, 52], ophthalmic slit-lamp biomicroscopes [3],catheters [26], endoscopes [29], laparoscopes [27], needles [28], and hand-heldimaging probes [27] (Fig. 8.6). Imaging penetration is determined by the opti-cal absorption and scattering properties of the tissue or specimen. The imagingpenetration for OCT ranges from tens of millimeters for transparent tissuessuch as the eye to less than 3 mm in highly scattering tissues such as skin.To image highly scattering tissues deep within the body, novel beam-deliveryinstruments have been developed to relay the OCT beam to the site of thetissue to be imaged. An OCT catheter has been developed for insertion intobiological lumens such as the gastrointestinal tract [29]. Used in conjunction

136 S.A. Boppart

Fig. 8.6. Beam delivery instruments. The OCT beam can be delivered through anumber of new or modified optical instruments including (a) surgical and researchmicroscopes, (b) hand-held probes and laparoscopes, (c) fiber-optic catheters, and(d) optical needle-probes

with endoscopy, the 1 mm diameter catheter can be inserted through the work-ing channel of the endoscope for simultaneous OCT and video imaging [31].Similar catheters have been used to image plaques within the living humancoronary arteries [33]. Minimally invasive surgical procedures utilize laparo-scopes, which are long, thin, rigid optical instruments to permit video-basedimaging within the abdominal cavity. Laparoscopic OCT imaging has beendemonstrated by passing the OCT beam through the optical elements of alaparoscope [27,32]. Deep solid-tissue imaging is possible with the use of fiber-needle probes [28]. Small (400 µm diameter) needles housing a single opticalfiber and micro-optic elements have been inserted into solid tissues and ro-tated to acquire OCT images. Recently, microfabricated micro-electro-optical-mechanical systems (MEOMS) technology has been used to miniaturize theOCT beam scan mechanism [53].

8.6 Spectroscopic Optical Coherence Tomography

Spatially distributed spectroscopic information can be extracted from the tis-sue specimen using spectroscopic OCT (SOCT) algorithms and techniques[54–57] (Fig. 8.7). In structural OCT imaging, the amplitude of the envelopeof the field autocorrelation is acquired and used to construct an image based onthe magnitude of the optical backscatter at each position. Spectral information


Fig. 8.7. Spectroscopic OCT imaging. Spatially resolved spectroscopic informationcan be extracted from the detected OCT signal. By digitizing the interferogramfringes and transforming the data using a Fourier-transform, the spectrum of thereturned light can be determined. When compared to the original laser spectrum,spectral differences can be identified, which are related to the absorption and scat-tering of the incident light. Spectroscopic OCT images of control and experimentalbotanical specimens where a highly absorbing and fluorescent chemical dye hasaccumulated in the vascular system. Corresponding fluorescence and bright-fieldmicroscopy images are shown. Figure modified with permission from [58]

can be obtained by digitizing the full interference signal and applying digitalsignal processing algorithms to transform the data from the time (spatial) do-main to the frequency (spectral) domain. Transformation algorithms includethe Morlet wavelet and the short-time Fourier transform with attention toreduction of windowing artifacts. While spectroscopic data can be extractedfrom each point within the specimen, there is an inherent trade-off betweenhigh spatial resolution and high spectral resolution and novel algorithms arebeing investigated to optimize this data. Once the spectral data is obtainedat a point in the tissue, the spectral center of mass can be calculated andcompared with the original spectrum from the laser source. Spectral shifts tolonger or shorter wavelengths, from the original center of mass, are displayedon a 2D image using a multidimensional hue-saturation-luminance (HSL) colorspace. At localized regions within the tissue, a color is assigned with a huethat varies according to the direction of the spectral shift (longer or shorterwavelength) and a saturation that corresponds to the magnitude of that shift.The luminance is held constant.

Longer wavelengths of light are scattered less in turbid media. In a homo-geneously scattering sample, one can observe using spectroscopic OCT thatshorter wavelengths are scattered near the surface and a smooth color-shiftoccurs with increasing depth, as longer wavelengths are scattered. In moreheterogeneous samples, such as tissue, scattering objects such as cells andsub-cellular organelles produce variations in the spectroscopic OCT data.

138 S.A. Boppart

Spectroscopic OCT images can indicate changes in the spectroscopic prop-erties of the tissue; however, further investigation is needed to determine howthe spectroscopic variations relate to the biological structures and how thisinformation can be used for diagnostic purposes.

8.7 Applications to Cancer Imaging

The noninvasive, noncontact, high-resolution, real-time imaging capabilitiesof OCT and its many modes of operation have enabled a wide range of newapplications in biology, medicine, surgery, and materials investigations. In thissection, applications in tumor cell biology and cancer imaging are presented.OCT has successfully made a transition from being a laboratory-based exper-imental technology to one that is useful clinically [59]. In the coming years,results from long-term controlled clinical trials will further demonstrate theusefulness of this technology.

8.7.1 Cellular Imaging for Tumor Cell Biology

Although previous studies have demonstrated in vivo OCT imaging of tissuemorphology, most have imaged tissue at ∼10–15 µm resolutions, which doesnot allow differentiation of cellular structure. The ability of OCT to identifythe mitotic activity, the nuclear-to-cytoplasmic ratio, and the migration ofcells has the potential to not only impact the fields of cell, tumor, and de-velopmental biology, but also impact medical and surgical disciplines for theearly diagnostics of disease such as cancer.

High-resolution in vivo cellular and subcellular imaging has been demon-strated in the Xenopus laevis (African frog) tadpole (Fig. 8.8) [16, 60]. Manyof the cells in this common developmental biology animal model are rapidlydividing and migrating during the early growth stages of the tadpole, provid-ing an opportunity to image dynamic cellular processes. Cell dynamics canalso be tracked in three-dimensional volumes of high-resolution OCT data asthey migrate through an engineered tissue scaffold along a chemoattractant-induced gradient (Fig. 8.9) [61]. From this 3D data set, time-dependent cellposition was color-coded. The ability of OCT to characterize cellular dynam-ics such as mitosis and migration is relevant for cancer diagnostics and for theinvestigation of tumor metastasis in humans.

Combining the coherence gating of OCT with high numerical aperture mi-croscope objectives enables high axial- and transverse-resolution imaging deepwithin highly scattering specimens. This optical configuration has been calledoptical coherence microscopy (OCM), and can improve the optical sectioningcapability of confocal microscopy. For several years, investigators have recog-nized that OCT and MPM can utilize a single laser source for multimodal-ity imaging. Recently, a microscope that integrates OCM and multiphoton(MPM) fluorescence imaging has been used to image cells in 3D engineered


Fig. 8.8. In vivo morphological and cellular imaging in a developmental biologyanimal model. Photograph and cross-sectional OCT images of a Xenopus laevis(African frog) tadpole specimen showing developing tissue morphology, as well asindividual cells and corresponding histology

Fig. 8.9. Tracking cell migration. A population of macrophage cells is tracked inthree dimensions over the course of 3 h using OCT. Macrophage cell migration wasinduced with a chemoattractant at one end of the 3D scaffold, separated by a semi-permeable membrane. The time-dependent positions of the cells are color-codedwhite (time = 0h) and grey (time = 3h). Modified figure used with permissionfrom [61]

tissue scaffolds [62, 63]. OCT and MPM provide complementary image data.OCT can image deep through transparent and highly scattering structuresto reveal the 3D structural information. OCT, however, cannot detect thepresence of a fluorescing particle. In a complementary manner, MPM can lo-calize fluorescent probes in three-dimensional space and provide insight intocell function. MPM can detect the fluorescence, but not the microstructure

140 S.A. Boppart

Fig. 8.10. Integrated multimodality microscopy using optical coherence microscopy(OCM) and multiphoton microscopy (MPM). Coregistered image data providesinsight into structure (OCM) and function (MPM) of cells in 2D and 3D culture con-ditions, under the influence of external mechanical stimuli, and following pharmaco-logical interventions. Fibroblasts with green-fluorescent-protein (GFP) transfectedto be expressed with vinculin, a cell surface adhesion protein, are shown. A nuclearfluorescent dye has also been used to identify the location of the nuclei. OCM im-ages based on scattering show the spatial distribution of the cell and extracellularcomponents. Figure modified and used with permission from [63]

nor the location of the fluorescence relative to the microstructure. Hence, thedevelopment of an integrated microscope capable of OCT and MPM uniquelyenables the simultaneous acquisition of microstructural data and the localiza-tion of fluorescent probes for precisely coregistered structural and functionalimaging (Fig. 8.10).

8.7.2 Translational Breast Cancer Imaging

OCT has been used to differentiate between the morphological structure ofnormal and neoplastic tissue for a wide-range of tumors [13, 64–67]. The useof OCT to identify tumors and tumor margins in situ will represent a signif-icant advancement for medical or image-guided surgical applications. Beamdelivery systems such as a compact and portable hand-held surgical probeor a modified surgical microscope permits OCT imaging within the surgical


field, while the main OCT instrument can be remotely located in the surgi-cal suite. The use of OCT in image-guided surgical procedures represents aparadigm shift for the surgical oncology community. While the surgical on-cologist typically resects macroscopic solid tumors, consideration must stillbe given to the microscopic, cellular extent of the disease. To address this,surgeons typically resect with large margins in an effort to remove any occultcells or nests of tumor. Resected tissue is sent to the surgical pathology labwhere margins are examined following histological processing to ensure thatthey are clean. During this time, a fully staffed operating room and anes-thetized patient must often wait for the decision on these margins. Therefore,moving the high-resolution microscopic imaging of tumor margins from thesurgical pathology lab into the operating room would represent a substantialreduction in costs, including time, costs, and patient health. The use of OCTfor examining tumor margins has many advantages, as well as introducingnew challenges. Large surface areas (and three-dimensional volumes) of tissuemust be examined in real-time at micron resolutions. Improvements in datamanagement, acquisition rates, automated tissue identification, and zoomingcapabilities all must be addressed.

An example of the use of OCT for identifying tumors at varying stages isshown in Fig. 8.11. Using a well-characterized carcinogen-induced rat mam-mary tumor model that mimics the progression and histological findings ofhuman ductal carcinoma of the breast, OCT images were acquired at varyingtime-points and compared to corresponding histology [67]. In good agreement,the OCT images identify not only late-stage morphological changes and theclear tumor margin, but also early ductal changes and evidence of abnormalcells located away from the primary tumor.

A portable OCT system has been constructed for intraoperative imaging ofsurgical margins following lumpectomy procedures for the surgical treatmentof breast cancer (Fig. 8.12). Different OCT biopsy needles have also been con-structed for imaging into solid tumors or for guiding needle-biopsy proceduresin breast cancer [28] (Fig. 8.12). Ongoing intraoperative and image-guidedprocedure studies are underway to determine the sensitivity and specificity ofOCT compared with the gold-standard histopathological analysis.

8.8 Optical Coherence Tomography Contrast Agents

When imaging biological tissues, it is often desirable to enhance the signalsmeasured from specific structures. Contrast agents that produce a specificimage signature have been utilized in virtually every imaging modalities, in-cluding ultrasound, computed tomography, magnetic resonance imaging, andoptical microscopy, among many others. There are multiple optical propertiesthat are amenable for generating contrast in optical images, including OCTimages (Fig. 8.13). Recently, new engineered contrast agents and molecularcontrast techniques specifically designed for OCT have been developed and

142 S.A. Boppart

Fig. 8.11. Cancer imaging using OCT. OCT can identify neoplastic changes atvarying stages of tumor growth. Top image set: Morphological OCT and histologicalimaging of (a,b) normal and (c,d) late-stage carcinogen-induced mammary tumor ina rat model. The tumor (t) mass is evident compared to the low-backscattering adi-pose (a) cells. Bottom image set: Early-stage ductal changes detected in this modelusing OCT and confirmed with histology. Ducts are imaged in cross-section (a,b)and through a longitudinal section (c,d). Figure modified and used with permissionfrom [67]

characterized [68, 69]. In contrast to fluorescent probes, which are commonlyused as contrast agents in fluorescence, confocal, and multiphoton microscopy,new classes of optical contrast agents suitable for OCT must be based on mech-anisms other than the detection of incoherently emitted fluorescence becauseOCT detects only coherent light. Agents that alter the local scattering orabsorption properties are used (Fig. 8.14), such as oil-filled microspheres thathave scattering nanoparticles of melanin, gold, carbon, or iron-oxide eitherembedded in their protein shell or encapsulated in their liquid-filled core [70].Fluorescence-based microscopy techniques have the advantage of very lowbackground signal in the absence of autofluorescence. This advantage can alsobe obtained in OCT by using novel dynamic contrast agents that are physicallymodulated in space using an external magnetic or electric field [71,72]. While


Fig. 8.12. Portable OCT system for real-time imaging during surgical operationsand procedures. The compact system utilizes a superluminescent diode with a centerwavelength at 1,300 nm wavelength and spectral-domain detection. (a) Forward-directed and (b) side-directed fiber-optic OCT needle probes can be used to collectdata from within solid tissue masses or during needle-biopsy procedures

the resolutions of OCT may be insufficient to resolve individual nanometer ormicron-sized scattering agents, the use of modulating particles enables sub-resolution detection of particles moving in and out of the OCT beam, or bymodulating not only the agent itself, but also the local tissue environment,producing both changes in the amplitude of the reflected light (for structuralOCT) and the phase of the reflected light (for phase-resolved spectroscopic orDoppler OCT). Because OCT detects changes in optical scattering, the sen-sitivity of detecting small relative changes in scattering is less than detectingrelatively large changes in local absorption, particularly if absorption is dueto an exogenously administered contrast agent. There is an increasing interestin near-infrared contrast agents in fluorescence-based microscopy techniquesbecause longer excitation and emission wavelengths can propagate deeper intoand further out of scattering tissue. Spectroscopic OCT techniques, which candetect spatially resolved spectral changes, can be used to identify the presenceof a near-infrared absorbing dye if the absorption spectra of the dye overlapswith the optical source spectrum [58].

144 S.A. Boppart

Fig. 8.13. Metaphorical diagram of an optical contrast agent illustrating the mul-tiple optical substrates, properties, and applications that can be utilized for opticalprobes and contrast agents. There are many properties and principles to exploit, inaddition to the more common fluorescence and bioluminescence properties. Figureused with permission from [68]

Fig. 8.14. Contrast agents for OCT. As with every other imaging modality, con-trast agents enhance the diagnostic ability of the modality and permit site-specificdetection of features. A wide array of novel contrast agents have recently been de-veloped including (top left) microspheres with scattering nanoparticles embedded inthe core or shell, (top right) magnetically-modulated agents of free or encapsulatediron-oxide, (bottom left) absorbing near-infrared dyes detected with spectroscopicOCT, and (bottom right) plasmon-resonant nanorods as scattering and absorbingagents


Finally, plasmon-resonant gold nanoparticles in the form of spheres, shells,cages, and rods have been utilized as strong as wavelength-specific absorbersand scatterers [73–75]. By changing the size and structure of these nanopar-ticles, the wavelength-dependent absorption and scattering can be tunedthroughout the near-infrared and visible wavelengths. These nanoparticleshave been used as contrast agents for OCT, and also proposed as multifunc-tional therapeutic agents because their strong absorption properties can beused to induce local hyperthermia in cells and tissues.

All of these agents are expected to be highly biocompatible and composedof materials that have been previously found to be suitable for in vivo use.These agents, with protein, iron-oxide, gold, or biocompatible molecular sur-faces, may be functionalized with antibodies or molecules to target them tospecific molecules, cells, or tissue types and thus provide additional selectivitythat can enhance the utility of OCT as an emerging diagnostic technique.

8.9 Molecular Imaging using Optical CoherenceTomography

The advantages of OCT, compared to other imaging techniques, are numer-ous. In particular, OCT can provide imaging resolutions that approach thoseof conventional histopathology and imaging can be performed in situ. Despiteits advantages, a serious drawback to OCT is that the linear scattering prop-erties of pathological tissue probed by OCT are often morphologically and/oroptically similar to the scattering properties of normal tissue. For example,although morphological differences between normal and neoplastic tissue maybe obvious at later tumor stages, it is frequently difficult to optically detectearly-stage tumors. To improve the ability of discriminating tissue types inthis scenario, molecular imaging techniques are being developed.

Molecular imaging involves the generation of images or maps of tissuethat contain molecular-specific information. The use of exogenous contrastagents in OCT (described above) is one example of molecular imaging, whensuch agents are functionalized to target and label specific molecular structureson cells or in tissue, such as cell-surface receptors that are over-expressed intumors. In many instances, it may be more desirable to detect the presence ofendogenous molecules without the addition exogenous agents that could alterthe biology or the viability of the tissue. Pump and probe techniques have beenused to detect the presence of molecules that have transient absorption statesin the tissue that are induced by an external pump beam [76]. SpectroscopicOCT has been used to probe for variations in the oxygenation state of tissue,recognizing the spectroscopic differences between oxy- and deoxy-hemoglobin[77]. Both of these methods, however, rely on a limited set of molecules ormolecular features that can be detected.

A new method called Nonlinear Interferometric Vibrational Imaging(NIVI) has been developed to achieve molecular contrast in OCT imaging by

146 S.A. Boppart

exploiting optical nonlinearities [78, 79]. The nonlinear processes, in particu-lar, are coherent anti-Stokes Raman scattering (CARS) and second harmonicgeneration (SHG), but the general idea could easily be extended to other non-linear processes such as third harmonic generation (THG), coherent StokesRaman scattering (CSRS). For these, the nonlinear process generates an op-tical signal that remains coherent with the incident light. Therefore in NIVI,the origin of this coherent signal can be spatially resolved in three-dimensionsusing OCT-like interferometers. The heterodyne detection scheme commonin OCT also imparts significant advantages in detecting these nonlinearsignals, including increased sensitivity, full reconstruction of the magnitudeand phase information of the sample Raman susceptibility [80], backgroundrejection of noncoherent four-wave-mixing processes arising from moleculessuch as water [81], and in the case where reference molecules are placed inthe reference arm of the interferometer, molecular specificity can be obtained.Figure 8.15 shows a schematic of the energy-level diagram involved in CARS,along with how a reference molecular species would be incorporated into

np = pump laser frequency

TunableSources

Delay (µm)−20 0 20

2.0

1.5

1.0

0.5

0.0

40 60

Lc = 32 µm

λas = 647 µm

80

2

0

Delay (nm)

I (a.

u.)

I CA

RS (

a.u.

)

0 647

Detector

Dm

Sample

Molecular/ChemicalSpecies

ReferenceMirror

BS

np

nsnp

nAS

ns = Stokes laser frequency

1>

0>

nAS = anti-Stokes output frequency

||

Fig. 8.15. Molecular OCT imaging. Nonlinear optical signatures in tissue are usedto identify specific molecular bonds in a technique called Nonlinear InterferometricVibrational Imaging (NIVI). An energy level diagram is shown (top left) for coherentanti-Stoke Raman scattering (CARS). By placing a reference molecular species in thereference arm of the OCT interferometer (top right), coherent CARS signals are gen-erated in the reference molecular species, and from the sample to produce spatiallyresolved images of molecular composition. An interferogram from two independentlygenerated CARS beams is shown (bottom) for the benzene molecule, demonstratingthe familiar-looking OCT interferogram, but with molecular specificity


0

200

2000

1500

1000

500

0

3000 4000 5000 6000 7000

Micrometers

(a)

(b) (c)

Meniscus

Nonlinear InterferometricVibrational Imaging

TransmissionLight Microscopy

Mic

rom

eter

sM

icro

met

ers

8000 9000 10000

1000 µm

11000 12000

400

600

800

1000

1200

1400

1600

Fig. 8.16. Multidimensional Nonlinear Interferometric Vibrational Imaging (NIVI).NIVI, tuned to the C–H vibrational resonance, is performed on a cuvette containingacetone. Signal is generated only from the acetone and not from the cuvette or theparticulates in the acetone. NIVI images are shown for the meniscus and bottomof the cuvette, and compared with a transmission light microscopy image of thecuvette bottom. Modified figures used with permission from [82]

the OCT interferometer. A representative interferogram generated from twoindependent samples of benzene is shown, demonstrating that a molecularlyspecific interferogram can be generated, and used to enhance molecular con-trast and perform molecular imaging using NIVI [78]. This has been extendedto multidimensional imaging (Fig. 8.16), and on-going research is investigat-ing NIVI for 3D molecular imaging in tissues. While this method is still indevelopment, the ability to image and map the three-dimensional spatialdistribution of individual endogenous molecules such as DNA, RNA, pro-teins, or other diagnostic biomolecules [79] is likely to extend the diagnosticcapabilities and clinical utility of OCT to the molecular level.

8.10 Conclusions

The capabilities of OCT offer a unique and informative means of imaging bi-ological specimens and nonbiological samples. The noncontact nature of OCTand the use of low-power near-infrared radiation for imaging allow this tech-nique to be noninvasive and safe. Conventional structural OCT imaging doesnot require the addition of fluorophores, dyes, or stains to provide contrast

148 S.A. Boppart

in images under most conditions. Instead, OCT relies on the inherent opticalcontrast generated from variations in optical scattering and index of refrac-tion. These factors permit the repeated use of OCT for extended imaging.OCT permits the cross-sectional imaging or en face sectioning of tissue andsamples, enabling in vivo structure to be visualized in opaque specimens, orin specimens too large for high-resolution confocal or light microscopy.

Imaging at cellular and subcellular resolutions with OCT is an importantarea of ongoing research. While many developmental biology animal modelshave been commonly used for their scientific value, small size, ease of care andhandling, and readily visualized cellular features, cellular imaging in humans,particularly in situ, is a challenge because of the smaller cell sizes (10–20 µm)compared to larger undifferentiated cells in developing organisms. To our ad-vantage, poorly differentiated cells present in many neoplastic tissues tend tobe larger, increasing the likelihood for detection using OCT at current imagingresolutions.

Clinical applications of OCT are likely to continue to increase, particularlyas more commercial OCT systems become available. With successful integra-tion into clinical ophthalmic imaging, OCT applications in gastroenterology,cardiology, and oncology are likely to become more commonplace. Finally,novel means to perform molecular imaging using OCT will be pursued, usingtargeted OCT-specific contrast agents or advanced nonlinear optical methodsfor identifying and mapping the molecular and ultrastructural composition ofbiological tissue. OCT represents a multifunctional investigative tool, whichnot only complements many of the existing imaging technologies availabletoday, but overtime, is also likely to become established as a major opticalimaging modality.

Acknowledgments

I thank my students, post-doctoral fellows, and research personnel in theBiophotonics Imaging Laboratory at the Beckman Institute for their hardwork and dedication in advancing the OCT technology. I appreciate the insightand collaborative contributions of Profs. Kenneth Suslick, Martin Gruebele,and Keith Singletary from the University of Illinois at Urbana-Champaign andAlexander Wei from Purdue University for using OCT to explore new applica-tions. I also thank my colleagues in Biomedical Optics for their contributionsto this work, and my clinical collaborators and colleagues at Carle Founda-tion Hospital and Carle Clinic Association, Urbana, Illinois, USA. Additionalinformation can be obtained at http://biophotonics.uiuc.edu. Prof. StephenBoppart’s email address is [email protected].


References

1. D. Huang, E.A. Swanson, C.P. Lin et al., Science 254, 1178 (1991)2. B.E. Bouma, G.J. Tearney (eds.), Handbook of Optical Coherence Tomography

(Marcel Dekker, New York, 2001)3. J.S. Schuman, C.A. Puliafito, J.G. Fujimoto, Optical Coherence Tomography of

Ocular Diseases, 2nd edn. (Slack, Thorofare, NJ, 2004)4. J.M. Schmitt, A. Knuttel, M. Yadlowsky et al., Phys. Med. Biol. 39, 1705 (1994)5. J.G. Fujimoto, C. Pitris, S.A. Boppart et al., Neoplasia 2, 9 (2000)6. M.E. Brezinski, G.J. Tearney, B.E. Bouma et al., Circulation 93, 1206 (1996)7. G.J. Tearney, M.E. Brezinski, S.A. Boppart et al., Circulation 94, 3013 (1996)8. I.E. Jang, B.E. Bouma, D.H. Kang et al., J. Am. Coll. Cardiol. 39, 604 (2002)9. G.J. Tearney, M.E. Brezinski, J.F. Southern et al., Am. J. Gastroenterol. 92,

1800 (1997)10. J.A. Izatt, M.D. Kulkarni, H.W. Wang et al., IEEE J. Sel. Top. Quant. Electron.

2, 1017 (1996)11. G.J. Tearney, M.E. Brezinski, J.F. Southern et al., J. Urol. 157, 1915 (1997)12. E.V. Zagaynova, O.S. Streltsova, N.D. Gladkova et al., J. Urol. 167, 1492 (2002)13. S.A. Boppart, M.E. Brezinski, C. Pitris et al., Neurosurgery 43, 834 (1998)14. B.W. Colston Jr., M.J. Everett, L.B. Da Silva et al., Appl. Opt. 37, 3582 (1998)15. B. Bouma, G.J. Tearney, S.A. Boppart et al., Opt. Lett. 20, 1486 (1995)16. W. Drexler, U. Morgner, F.X. Kartner et al., Opt. Lett. 24, 1221 (1999)17. B. Povazay, K. Bizheva, A. Unterhuber et al., Opt. Lett. 27, 1800 (2002)18. H. Ren, T. Sun, D.J. MacDonald et al., Opt. Lett. 31, 927 (2006)19. E.C.W. Lee, J.F. de Boer, M. Mujat et al., Opt. Exp. 14, 4403 (2006)20. Z. Chen, T.E. Milner, S. Srinivas et al., Opt. Lett. 22, 1119 (1997)21. J.A. Izatt, M.D. Kulkarni, S. Yazdanfar et al., Opt. Lett. 22, 1439 (1997)22. J.K. Barton, A.J. Welch, J.A. Izatt, Opt. Exp. 3, 251 (1998)23. J.F. de Boer, T.E. Milner, M.J.C. van Germert et al., Opt. Lett. 22, 934 (1997)24. M.J. Everett, K. Schoenenberger, B.W. Colston Jr. et al., Opt. Lett. 23, 228

(1998)25. K. Schoenenberger, B.W. Colston Jr., D.J. Maitland et al., Appl. Opt. 37, 6026

(1998)26. G.J. Tearney, S.A. Boppart, B.E. Bouma et al., Opt. Lett. 21, 1 (1996)27. S.A. Boppart, B.E. Bouma, C. Pitris et al., Opt. Lett. 22, 1618 (1997)28. A.M. Zysk, S.G. Adie, J.J. Armstrong et al., Opt. Lett. 32, 385 (2007)29. G.J. Tearney, M.E. Brezinski, B.E. Bouma et al., Science 276, 2037 (1997)30. A. Das, M.V. Sivak Jr., A. Chak et al., Gastrointest. Endosc. 54, 219 (2001)31. X.D. Li, S.A. Boppart, J. Van Dam et al., Endoscopy 32, 921 (2001)32. F.I. Feidchtein, G.V. Gelikonov, V.M. Gelikonov et al., Opt. Exp. 3, 257 (1998)33. H. Yabushita, B.E. Bouma, S.L. Houser et al., Circulation 106, 1640 (2002)34. K. Takada, I. Yokohama, K. Chida et al., Appl. Opt. 26, 1603 (1987)35. A.F. Fercher, K. Mengedoht, W. Werner, Opt. Lett. 13, 186 (1988)36. C.K. Hitzenberger, Invest. Ophthalmol. Vis. Sci. 32, 616 (1991)37. A.M. Rollins, J.A. Izatt, Opt. Lett. 24, 1484 (1999)38. B.E. Bouma, G.J. Tearney, Opt. Lett. 4, 531 (1999)39. M.A. Chroma, M.V. Sarunic, C. Yang et al., Opt. Exp. 11, 2183 (2003)40. B. Cense, N.A. Nassif, T.C. Chen et al., Opt. Exp. 12, 2435 (2004)41. S.H. Yun, G.J. Tearney, J.F. de Boer et al., Opt. Exp. 11, 2953 (2003)

150 S.A. Boppart

42. R. Huber, M. Wojtkowski, J.G. Fujimoto, Opt. Exp. 14, 3225 (2006)43. H. Okamoto, M. Wada, Y. Sakai et al., J. Lightwave Tech. 16, 1881 (1998)44. B.E. Bouma, L.E. Nelson, G.J. Tearney et al., J. Biomed. Opt. 3, 76 (1998)45. M. Bashkansky, M.D. Duncan, L. Goldberg et al., Opt. Exp. 3, 305 (1998)46. I. Hartl, X.D. Li, C. Chudoba et al., Opt. Lett. 26, 608 (2001)47. D.L. Marks, A.L. Oldenburg, J.J. Reynolds et al., Opt. Lett. 27, 2010 (2002)48. Y. Wang, Y. Zhao, J.S. Nelson et al., Opt. Lett. 28, 182 (2003)49. W. Denk, J.H. Strickler, W.W. Webb, Science 248, 73 (1990)50. E. Beaurepaire, L. Moreaux, F. Amblard et al., Opt. Lett. 24, 969 (1999)51. S.A. Boppart, G.J. Tearney, B.E. Bouma et al., Proc. Nat. Acad. Sci. USA 94,

4256 (1997)52. A.G. Podoleanu, J.A. Rogers, D.A. Jackson et al., Opt. Exp. 7, 292 (2000)53. Y. Pan, H. Xie, G.K. Fedder, Opt. Lett. 26, 1966 (2001)54. U. Morgner, W. Drexler, F.X. Kartner et al., Opt. Lett. 25, 111 (2000)55. R. Leitgeb, M. Wojtkowski, A. Kowalczyk et al., Opt. Lett. 25, 820 (2000)56. C. Xu, S.A. Boppart, Opt. Exp. 12, 4790 (2004)57. C. Xu, P.S. Carney, S.A. Boppart, Opt. Exp. 13, 5450 (2005)58. C. Xu, J. Ye, D.L. Marks et al., Opt. Lett. 29, 1647 (2004)59. A.M. Zysk, F.T. Nguyen, A.L. Oldenburg et al., J. Biomed. Opt. 12(5), 051403

(2007)60. S.A. Boppart, B.E. Bouma, C. Pitris et al., Nat. Med. 4, 861 (1998)61. W. Tan, A.L. Oldenburg, J.J. Norman et al., Opt. Exp. 14, 7159 (2006)62. C. Vinegoni, T.S. Ralston, W. Tan et al., Appl. Phys. Lett. 88, 053901 (2006)63. C. Xu, C. Vinegoni, T.S. Ralston et al., Opt. Lett. 31, 1079 (2006)64. A.M. Sergeev, V.M. Gelikonov, G.V. Gelikonov et al., Opt. Exp. 1, 432 (1997)65. S.A. Boppart, A.K. Goodman, C. Pitris et al., Br. J. Obstet. Gynaecol. 106,

1071 (1999)66. Y.T. Pan, T.Q. Xie, C.W. Du et al., Opt. Lett. 28, 2485 (2003)67. S.A. Boppart, W. Luo, D.L. Marks et al., Breast Cancer Res. Treat. 84, 85

(2004)68. S.A. Boppart, A.L. Oldenburg, C. Xu et al., J. Biomed. Opt. 10, 041208 (2005)69. C. Yang, Photochem. Photobiol. 81, 215 (2005)70. T.M. Lee, A.L. Oldenburg, S. Sitafalwalla et al., Opt. Lett. 28, 1546 (2003)71. A.L. Oldenburg, J.R. Gunther, S.A. Boppart, Opt. Lett. 30, 747 (2005)72. A.L. Oldenburg, F.J. Toublan, K.S. Suslick et al., Opt. Exp. 13, 6597 (2005)73. A.L. Oldenburg, M.N. Hansen, D.A. Zweifel et al., Opt. Exp. 14, 6724 (2006)74. C. Loo, A. Lin, L. Hirsch et al., Technol. Cancer Res. Treat. 3, 33 (2004)75. J. Chen, F. Saeki, B.J. Wiley et al., Nano Lett. 5, 473 (2005)76. K.D. Rao, M.A. Choma, S. Yazdanfar et al., Opt. Lett. 28, 340 (2003)77. D.J. Faber, E.G. Mik, M.C.G. Aalders et al., Opt. Lett. 28, 1436 (2003)78. C. Vinegoni, J.S. Bredfeldt, D.L. Marks et al., Opt. Exp. 12, 331 (2004)79. D.L. Marks, S.A. Boppart, Phys. Rev. Lett. 92, 123905 (2004)80. G.W. Jones, D.L. Marks, C. Vinegoni et al., Opt. Lett. 31, 1543 (2006)81. D.L. Marks, C. Vinegoni, J.S. Bredfeldt et al., Appl. Phys. Lett. 85, 5787 (2004)82. D.L. Marks, C. Vinegoni, J.S. Bredfeldt et al., Opt. Photon. News 12, 23 (2005)

9

Coherent Laser Measurement Techniquesfor Medical Diagnostics

B. Kemper and G. von Bally

9.1 Introduction

Holography and speckle interferometry are well established tools for industrialnondestructive testing and quality control, which are in general performed bythe detection of object displacements while applying static stress, tempera-ture changes, shock waves, or by vibration monitoring [1–3]. Therefore, upto the present, various holography and speckle interferometry systems formacroscopic as well as microscopic applications have been developed. Alsofor biomedical applications holographic and speckle interferometric metrologyopens up new perspectives for the visualization and the detection of displace-ments and movements. Here, for the early recognition of malignant cancers,for example, it is of interest to distinguish between different tissue elastic-ities [4–7]. Also, in the fields of Life Sciences and Biophotonics, there is arequirement for an optical instrumentation of timely and spatially high res-olution analysis, measurement, and documentation in supracellular, cellular,and subcellular range. For such applications, the special requirements of tis-sues and cellular probes have to be taken into account and a lateral resolutionin microscopic dimensions as well as very compact and flexible speckle inter-ferometric arrangements are required. Additionally, for in vivo applications,a high repetition rate of the results for online monitoring is necessary. Thiscan be obtained by the application of micro-optics, optical fiber technology,miniaturized (color) CCD sensors, and fast digital image processing systems.Furthermore, spatial phase shifting methods and suitable techniques for non-diffractive digital holographic reconstruction, which require in contrast to tem-poral techniques considerably less stability of the experimental setup and theinvestigated specimens, allow for a quantitative determination of the objectwave phase. From this information the underlying displacement data as wellas the object shape can be obtained under suitable geometric conditions [1,8].In this contribution, in an overview, experimental results obtained with sev-eral holographic and (speckle) interferometric arrangements based on endo-scope and microscope optics are presented to demonstrate applications and

152 B. Kemper and G. von Bally

possibilities of holographic interferometric metrology on biological specimensand cellular samples. The results of these investigations show that it is possi-ble to detect minimal-invasive elasticity differences of biological tissues, whichmay be used for an early tumor and cancer recognition. Furthermore, dataobtained from long-term digital holographic investigations on toxin-inducedreactions of adherent cancer cells demonstrate application prospects of digitalholographic phase contrast microscopy in marker-free life cell imaging [9,10].

9.2 Electronic Speckle Pattern Interferometry (ESPI)

9.2.1 Double Exposure Subtraction ESPI

Figure 9.1 shows schematically a mainly out-of-plane sensitive ESPI for theinvestigation of rough surfaces. Coherent laser light is divided into an objectillumination wave O and a reference wave R. The image of the object surfacethat is covered by a speckle pattern due to the speckle effect [11] is imagedon an image recording device (e.g., a charge coupled device (CCD) sensor).The superimposition of O and R is performed by a second beam splitter BS.In double exposure subtraction ESPI, two speckle patterns are recorded sub-sequently on a CCD sensor superimposed by a reference wave. The intensitiesI and I ′ of the two object states are [1]

I(x, y) = IO(x, y) + IR(x, y) + 2√

IO(x, y)IR(x, y) cos(ψ(x, y))

I ′(x, y) = IO(x, y) + IR(x, y) (9.1)

+ 2√

IO(x, y)IR(x, y) cos(∆ϕ(x, y) + ∆φ(x, y)).

Fig. 9.1. ESPI setup. BS beam splitter; L1, L2, L3 lenses; CCD CCD sensor

9 Coherent Laser Measurement Techniques for Medical Diagnostics 153

In (9.1) IO represents the intensity of the object wave, which can be as-sumed equal before and after the object deformation or movement betweentwo recordings and IR is the intensity of the reference wave. The term∆ψ(x, y) = φO(x, y)−φR(x, y) represents the stochastic phase difference while∆φ(x, y) is the phase difference caused by the object deformation d and thesensitivity vector S = kQ + kB, which takes into account the illuminationdirection kQ, the observation direction kB, and the light wavelength λ [1]:

∆φ(x, y) =2π

λS · d. (9.2)

Subtraction of the two intensities (9.1) yields

(I − I ′)(x, y) = 4√

IO(x, y)IR(x, y) sin[ψ(x, y) +

∆φ(x, y)2

]sin

∆φ(x, y)2

.

(9.3)

To display the result in video repetition rate on a monitor the modulus |I−I ′|is calculated. By observation of |I−I ′| qualitative information about the objectdeformation is obtained.

9.2.2 Spatial Phase Shifting (SPS) ESPI

Investigations on biological specimen require a method for quantitative deter-mination of the phase difference ∆φ that works even under instable conditions.For this reason, spatial phase shifting (SPS) methods [12, 13] are applied forsign correct phase difference determination that require only a single inter-ferogram to determine the phase distribution of one object state. Anotheradvantage of this technique in comparison with temporal phase shifting pro-cedures [14] is that no movable parts (e.g., Piezo translators) in the opticalpath and no additional electronic synchronization are necessary.

For the realization of SPS, the speckle image of the investigated object sur-face is superimposed in the image plane with spatial carrier fringes, e.g., withvertical or horizontal orientation. This is achieved by an adequately chosen,almost constant spatial phase gradient β(x, y) = (βx, βy) that is generated bya tilt between object wave and reference wave.

The intensity of the interference pattern formed on the CCD sensor canbe described as [14]

I(xm, yn) = I0(xm, yn)[1 + γ(xm, yn) sinc

(Φ

2

)cos(φ(xm, yn) + mβx + nβy + C)

].

(9.4)


In (9.4) m = 1, ...,M and n = 1, ..., N are the indices of the CCD pixelco-ordinates xm, yn. I0(xm, yn) = I0

O(xm, yn) + I0R(xm, yn) is the sum of the

intensities of object wave and reference wave. The parameter γ(xm, yn) =2√

IO(xm, yn)IR(xm, yn)/I0(xm, yn) denotes the modulation of the intensitypatterns. The term sinc(Φ/2) = sin (Φ/2)/(Φ/2) describes the intensity inte-gration on a single CCD pixel in the angle range Φ. φ(xm, yn) is the distrib-ution of the object wave phase and C an additional constant phase off-set.

For the calculation of the object wave phase from each recorded intensitypattern, in addition to Fourier transformation methods [15–17], a variablethree-step algorithm [12] can be applied by taking into account three intensityvalues Ik−1, Ik, Ik+1 (in horizontal or vertical orientation) of neighboring CCDpixels:

φk + kβ + C = arctan(

1 − cos β

sin β

Ik−1 − Ik+1

2Ik − Ik−1 − Ik+1

)modulo 2π.

(9.5)

The algorithm requires for a correct phase evaluation that the mean specklesize is at least the size of three pixels of the digitized interferograms. For theparameter β, either βx or βy may be chosen. The adjustment of the phasegradients can be performed by analysis of the 2D frequency spectrum of thecarrier fringe pattern in the interferograms by 2D digital fast Fourier transform(FFT) [13,16].

The phase difference ∆φk modulo 2π between two phase states φk, φ′k of

the object wave front is calculated:

∆φk = (φ′k + kβ + C) − (φk + kβ + C) = φ′

k − φk. (9.6)

Figure 9.2 illustrates the evaluation process of spatial phase-shifted inter-ferograms for displacement detection. Figure 9.2a,b shows the spatial phase-shifted interferograms I, I ′ obtained from two different displacement states ofa tilted metal plate. In Fig. 9.2c,d the corresponding phase distributions φ, φ′

(mod 2π), calculated by (9.5) from Fig. 9.2a,b, are depicted. The correspond-ing correlation fringe pattern calculated by (9.3) is depicted in Fig. 9.2e. Theresulting phase difference distribution ∆φ mod 2π obtained by subtraction ofFig. 9.2c,d modulo 2π as well as the according filtered phase difference distri-bution is shown in Fig. 9.2f,g. Figure 9.2h,i finally represents the unwrappedphase difference after removal of the 2π ambiguity and the related pseudo3D representation of the data. From the data in Fig. 9.2h the underlying dis-placement of the plate is calculated by taking into consideration recordingand imaging geometry of the experimental setup by (9.2).


Fig. 9.2. Evaluation of spatial phase-shifted interferograms. (a),(b) Spatial phase-shifted interferograms I, I ′ of two displacement states of a tilted metal plate; (c),(d)Phase distributions φ, φ′ (mod 2π) calculated by (9.5) from (a) and (b); (e) Cor-relation fringe pattern calculated by (9.3); (f),(g) Raw and filtered phase differencedistribution mod 2π obtained by subtraction of (c) and (d) modulo 2π; (h),(i) Un-wrapped phase difference and corresponding pseudo 3D representation of ∆φ


9.3 Endoscopic Electronic Speckle PatternInterferometry (ESPI)

Endoscopy is a wide spread intracavity observation technique routinely used inminimal invasive diagnostics and industrial intracavity inspection. The com-bination of endoscopic imaging with speckle interferometric metrology allowsthe development of tools for a nondestructive quantitative detection of defectswithin body cavities, including the analysis of shape, structure, displacements,and vibrations of the object. In this way Electronic-Speckle-Pattern Interfer-ometry (ESPI) opens up new perspectives for biomedical applications, espe-cially in minimally invasive diagnostics in the medical field [18]. Contrary toearlier attempts in holographic endoscopy where single interferograms withQ-switched lasers were recorded [19], an on-line process analysis can be per-formed with a rate near video repetition frequency. This can be achieved byendoscopic ESPI systems with an external (proximal) interferometric arrange-ment using standard endoscope optics [4,20] or by an arrangement where theESPI system is positioned in the endoscope tip [21] (distal arrangement).Such endoscope ESPI systems open up the possibility to replace the opera-tor’s tactile sense, which is lost in endoscopic surgery, by visual information(“endoscopic taction”) [4, 5, 22].

9.3.1 Proximal Endoscopic ESPI

The setup for proximal endoscopic ESPI with the speckle interferometer po-sitioned outside the cavity is illustrated in Fig. 9.3. The advantage of such

Fig. 9.3. Arrangement for proximal endoscopic ESPI. CCD CCD sensor, AOMacousto-optic modulator, DG double pulse generator, BV/PC digital image process-ing system


an arrangement is that standard CCD cameras can be used. Additionally,it is possible to form modular concepts of the ESPI system with flexible aswell as rigid standard endoscopes for imaging. The light source is a cw laser(e.g., an argon-ion laser, λ = 514.5 nm or a frequency doubled Nd:YAG laser,λ = 532 nm). The regulation of the light intensity is performed by an acousto-optic modulator (AOM) using the first diffraction order. Behind the AOM,the laser beam is divided into the object illumination wave and the referencewave, with both coupled into single-mode optical fibers. Spatial phase shift-ing (SPS) (see Sect. 9.2.2) is applied to obtain quantitative information aboutdisplacements and motions. Therefore, an interference pattern between objectwave and reference wave is generated by positioning the end of the referencewave fiber with an off-set to the optical axis. The interference patterns of thesuperimposed object wave and reference wave are recorded by a progressive-scan (full frame) camera with a CCD sensor. A fast digital image processingsystem (BV/PC) allows an image acquisition of the intensity patterns up to25 Hz. Simultaneously, the corresponding correlation fringe patterns obtainedby image subtraction are displayed online with a rate of 12.5 Hz on an ex-ternal monitor. The exposure time and the delay between two exposures arevariable in order to be suitable for the measurement conditions of biologicalspecimen (for further details see [4,13,23]). Figure 9.4 shows on the left panela mobile endoscope ESPI system based on an analog CCD camera and animage processing system with frame grabber interface. On the right panelof Fig. 9.4 a modular endoscope ESPI system with IEEE1394 (“FireWire”)technology consisting of a laser unit, a speckle interferometric module, and anotebook-computer for image processing is depicted.

One of the current limitations of endoscopic imaging in the gastrointesti-nal tract is the absence of tactile perception, making it impossible to assessgastrointestinal wall elasticity during routine endoscopic examinations. Forthis reason it is of interest to distinguish between tissue elasticity for the

Fig. 9.4. (Left) Mobile endoscope ESPI system based on an analoge CCD cameraand an image processing system with frame grabber interface. (Right) Modularendoscope ESPI system with IEEE1394 (“FireWire”) technology consisting of alaser unit, a speckle interferometric module, and a notebook-computer for imageprocessing


Fig. 9.5. In vitro investigations on a human intestinal specimen by stimulation withan endoscopic ultra sound tube. (a),(d) White light images of a tissue part withoutpathological findings and of the tumorous tissue; (b),(c) Phase difference distribu-tions (mod 2π); (c),(f) Pseudo 3D plots of the displacement along the marked crosssections in (b) and (e)

early recognition of malignant cancers, for example. In vitro investigationson intestinal specimens with carcinoma tissue have been carried out. For theexperiment, the tissue is stimulated by an endoscopic ultrasonic tube (singlepulse, amplitude: ≈200 µm). Afterwards, during the relaxation process of thetissue, series of stroboscopic phase difference distributions are captured.

Figure 9.5 depicts characteristic results. Figure 9.5a,d shows the white lightimages of an investigated area of the specimen without pathological findingsin comparison to a tumorous part of the tissue. In Fig. 9.5b,e exemplary re-sults of phase difference distributions mod 2π obtained during the relaxationprocess of the tissue are depicted. The corresponding calculated surface de-formation along the cross sections in Fig. 9.5b,e is plotted in Fig. 9.5c,f. Thephase difference shows concentric fringes for the tissue part without patho-logical findings (Fig. 9.5b) and a parallel fringe distribution for the hardenedtumorous tissue (Fig. 9.5e). Thus, the tissue part without pathological findingscan be distinguished from the tumorous tissue of diminished elasticity.

9.3.2 Distal Endoscopic ESPI

To avoid the phase instabilities of image fiber bundles and the aberrations ofstandard rod lens endoscope imaging systems in endoscopic ESPI, a compactrealization of the ESPI camera system in the endoscope tip (distal arrange-ment) is of particular advantage [21]. In addition, a smaller speckle size can be


Fig. 9.6. Concept of a distal endoscope ESPI sensor. CCD CCD sensor (1/4 or1/6 in.)

obtained by short optical path length distances in a compact interferometricalignment, thus achieving a higher lateral resolution up to microscopic dimen-sions. The concept of an out-of-plane sensitive distal ESPI sensor is illustratedin Fig. 9.6. The coherent light source is an amplitude modulated cw laser. Theobject illumination wave (O) as well as the reference wave (R) are guided bysingle-mode optical fibers. For the object illumination O as well as for thereference wave R additional microlens systems are applied that are fixed atthe end of each optical fiber.

To achieve an arrangement suitable for endoscopic applications, the ref-erence wave is redirected by 180◦ with a prism and a beam splitter. Theinterference patterns of the superimposed waves R and O are recorded bya one-chip color CCD sensor (e.g., 1/4 inch sensor or 1/6 inch sensor; pixelresolution, 752 × 582 pixels; maximum diameter, 8 or 4 mm). For the quan-titative displacement detection, an aperture is used to regulate the specklesize to three CCD pixels for the application of spatial phase shifting tech-niques [9, 23]. Therefore, a spatial carrier fringe pattern between O and R isgenerated by positioning the end of the reference wave fiber with a lateraloff-set out of the optical interferometer axis. Figure 9.7 depicts a photo of adistal endoscope ESPI sensor based on a 1/6 inch CCD sensor.

Figure 9.8 shows results of investigations with a distal endoscopic speckleinterferometer based on 1/4 inch color CCD sensor. In Fig. 9.8a the white


Fig. 9.7. Distal endoscope ESPI sensor based on a 1/6 in. CCD sensor

Fig. 9.8. Results obtained with a distal endoscopic speckle interferometer based ona 1/4 in. CCD sensor. (a) White light image of an USAF 1951 test chart; (b) Corre-sponding speckle image (λ = 514.5 nm), a lateral resolution of 8.7 µm is obtained; (c)Correlation fringe pattern of displacement measurements on a tilted white paintedmetal plate; (d) Corresponding filtered phase difference distributions


Fig. 9.9. Results of in vitro investigations on a human stomach: Filtered phasedifference distributions of a “healthy” area (a) and a region containing an adenocarcinoma (b)

light image of a USAF 1951 test chart positioned in a distance of 5 mm infront of the sensor is depicted. Figure 9.8b shows the corresponding speckleimage affected by illumination with coherent light (λ = 514.5 nm). Within thespeckle image a lateral resolution of 8.7 µm is obtained. Figure 9.8c,d presentsresults from a recorded series of displacement measurements with a whitepainted metal plate, which was tilted between two recordings to demonstratethe fringe resolution and quality. Shown are the correlation fringe pattern(Fig. 9.8c) and the respective filtered wrapped phase difference distributionsmodulo 2π (Fig. 9.8d) converted to 256 gray levels. The maximum displace-ment resolution of the system in this configuration can be estimated from thenoise of the phase difference map to λ/7. To demonstrate the performance ofthe distal ESPI sensor on biological specimens analogous to the experimentsdescribed in Sect. 9.3.1, investigations on tumorous human stomach gastricwall (in vitro) have been carried out.

In Fig. 9.9 the filtered wrapped phase difference distribution of a “healthy”part of the specimen is depicted (imaged area about 1 cm2), which has beenstimulated manually with a test needle. The results are concentric fringe pat-terns around the tip of the needle. In contrast, an area containing an adenocarcinoma shows a parallel fringe pattern (Fig. 9.9b), indicating that the tissueelasticity in this region is diminished, which is in agreement with the resultsfrom Sect. 9.3.1.

9.4 Microscopic (Speckle) Interferometry

In microscope ESPI, the motivation is to integrate the interferometric metrol-ogy into microscope systems to achieve an enhancement of these techniquesby an additional high resolution detection and visualization of movements,displacements [24, 25], and refractive index changes. Figure 9.10 shows the


Fig. 9.10. Setup for a microscope (speckle) interferometer. Light source: frequencydoubled cw Nd:YAG laser (λ = 532 nm); SMF: single-mode fibers for object illumi-nation and reference wave; the illumination of the sample can be performed optionalwith incident light or in transmission; AP: aperture for regulation of the speckle sizeand suppression of scattered light

schematic setup for a microscopic (speckle) interferometer. The object is im-aged by a microscope lens on a CCD sensor. The coherent object illuminationis performed either in transmission or with incident light by single mode opti-cal fibers. The speckle size (or in the case of smooth waves disturbing scatteredlight due to transparent specimens) is regulated by an additional aperture APthat is positioned behind the microscope lens. The detection of optical pathlength changes affected by micro changes of the sample can be performed bya spatial phase shifting method as described in Sect. 9.2.2. Therefore, the ref-erence wave is generated by an optical single mode fiber placed in the planeof AP that is positioned with an off-set from the optical axis of the interfer-ometer to produce a nearly constant phase gradient between object wave andreference wave.

Figure 9.11 shows results obtained by a microscope (speckle) interfer-ometer setup in combination with a ×10 magnification microscope lens. InFig. 9.11a the white light image of an USAF 1951 test chart is depicted.Figure 9.11b shows the corresponding speckle image effected by illumina-tion with coherent laser light (λ = 532 nm) with a lateral resolution of7.8 µm. In Fig. 9.11c–f results from displacement measurements on a dyedprobe of an intestine carcinoma which was tilted in the experiment are de-picted. Figure 9.11c shows the white light image of the investigated area.The correlation fringe pattern, the corresponding phase difference distribu-tion, and the filtered phase difference fringes are presented in Fig. 9.11d–f.


Fig. 9.11. Results of measurements obtained with a ×10 magnification microscopelens. (a) White light image of an USAF 1951 test chart; (b) Speckle image of (a) byillumination with coherent laser light (λ = 532 nm); (c) White light image of a dyedpart of intestine carcinoma fixed on a glass slide; (d) Correlation fringe pattern ofthe tilted specimen in (a); (e) Corresponding phase difference distribution modulo2π; (f) Filtered phase difference distribution


Fig. 9.12. (a) Phase difference distribution of optical path length changes dueto micro-changes and micro-movements of human liver tumor cells (HepG2) aftert = 75 min (×40 magnification immersion microscope lens). (b) Correlation fringepattern obtained with a ×10 magnification microscope lens during investigations ofa living water flea. For both measurements the illumination of the specimen wasperformed in transmission, λ = 532 nm

The results demonstrate that displacements can be detected while the lat-eral resolution of the speckle image is in microscopic dimensions. Figure 9.12ashows results obtained by application of the interferometric arrangement inFig. 9.10 in combination with a ×40 magnification immersion microscope lensfor monitoring microchanges of human liver tumor cells (HepG2) in cell culturemedium. The depicted phase difference distribution (illumination in transmis-sion, λ = 532 nm) due to optical path length changes corresponds to micro-movements and microchanges of the specimens after t = 75 min. Figure 9.12bshows a correlation fringe pattern obtained from the investigation of a livingwater flea utilizing a ×10 magnification microscope lens, which is an exem-plary result of a recorded series of 50 images and demonstrates the feature ofonline monitoring.

9.5 Digital Holographic Microscopy

9.5.1 Principle and Measurement Setup

In Life Sciences and Biophotonics, a quantitative minimal invasive analysis ofdynamic life processes with resolution in cellular and subcellular scale is ofparticular interest. In connection with microscopy, digital holography providescontact-less, marker-free, quantitative phase contrast imaging.

Digital holography is based on the classic holographic principle, with thedifference that the hologram recording is performed by a digital sensor, e.g.,a CCD or CMOS camera [26,27]. The subsequent reconstruction of the holo-graphic image that includes the information about the object wave is carried


Fig. 9.13. Schematics for digital holographic microscopy using incident light (left)and inverse transmission arrangements (right) [28]

out numerically with a computer. Digital holographic microscopy (DHM) per-mits nondestructive, marker-free, and quantitative high resolution full-fieldphase contrast imaging of transparent samples such as living cells [9, 10].

Figure 9.13 depicts the schematics of two “off-axis” setups for digital holo-graphic microscopy, which are particularly suitable for the integration intocommercial microscopy systems. The coherent light of a laser (e.g., a fre-quency doubled Nd:YAG laser, λ = 532 nm) is divided into object illumina-tion and reference wave, using singlemode optical fibers. The left panel ofFig. 9.13 shows an incident light illumination arrangement for the investiga-tion of reflective objects. The set-up on the right panel of Fig. 9.13 is designedto investigate transparent objects such as living cell cultures. In both casesthe coherent laser light for the illumination of the sample is coupled into theoptical path of the microscope’s condenser by a beam splitter. The referencewave is superimposed onto the light reflected or transmitted by the object bya second beam splitter with a slight tilt against the object wave front. Thus,“off-axis” holograms are generated and recorded by a CCD camera. Afterhologram acquisition, the data is transmitted by an IEEE1394 (“FireWire”)interface to a digital image processing system; thus bypassing the need forcost intensive frame grabber cards with hardware-specific software. This ap-proach provides the advantage that commercial standard microscope opticswith high numerical aperture (e.g., water immersion and oil immersion) canbe used in combination with an optimized (Koehler-like) illumination of thesample. Furthermore, such a modular integration of the additional opticalcomponents for digital holography does not restrict the conventional functionof the microscopy systems [28].


Fig. 9.14. (Left) Inverse microscope with P.A.L.M. MicroBeam and attached dig-ital holographic microscopy module. The illumination of the sample is performedin transmission mode (built in cooperation with P.A.L.M. Microlaser TechnologiesGmbH, Bernried, Germany). An incubator allows temperature stabilized long-terminvestigations of livings cells. (Right) Upright fluorescence microscope with digitalholographic microscopy module for illumination of the sample in transmission andreflection (built in cooperation with Carl Zeiss Jena GmbH, Germany) [28]

Figure 9.14 shows an inverse (left) and an upright (right) microscope withattached digital holographic microscopy modules. The typical hologram cap-turing time depends on the applied imaging device (here CCD cameras) andamounts within the range of 1 ms. The reconstruction rate of the digital holo-grams depends mainly on the capability of the applied image processing sys-tem and the size of the digitized holograms. With current computer systems(e.g., a Pentium IV 2.8 GHz) and the reconstruction algorithm described inSect. 9.5.2 at a frame size of 512× 512 pixels reconstruction rates up to ∼4 Hzcan be achieved.

9.5.2 Nondiffractive Reconstruction

The reconstruction of the digitally recorded holograms is performed numeri-cally with a computer. In general, Fresnel-transformation based digital holo-graphic reconstruction methods generate not only the information containedin the object wave but in addition the intensity of the reference wave (“zeroorder”) and a “twin image” [29]. Furthermore, the size of the reconstructedholographic image depends on the reconstruction distance to the hologramplane. Thus, in digital holographic microscopy, the reconstruction of thedigitally captured holograms is performed by the application of a non-diffractive reconstruction method (NDRM) [30–34]. The intensity distribu-tion IHP(x, y, z0) in the hologram plane HP, located at z = z0, is formed bythe interference of the object wave O(x, y, z = z0) and the reference waveR(x, y, z = z0):


IHP(x, y, z0) = O(x, y, z0)O∗(x, y, z0) + R(x, y, z0)R∗(x, y, z0)+O(x, y, z0)R∗(x, y, z0) + R(x, y, z0)O∗(x, y, z0)

= IO(x, y, z0) + IR(x, y, z0)

+ 2√

IO(x, y, z0)IR(x, y, z0) cos ∆ϕHP(x, y, z0), (9.7)

with IO = OO∗ = |O|2 and IO = RR∗ = |R|2 (* denotes the conjugate com-plex term). The parameter ∆ϕHP (x, y, z0) = φR(x, y, z0)− φO(x, y, z0) is thephase difference between O and R at z = z0. In the presence of a sample in theoptical path of O, the phase distribution represents the sum φO (x, y, z0) =φO0(x, y, z0)+∆ϕS(x, y, z0), where φO0(x, y, z0) denotes the pure object wavephase and ∆ϕS(x, y, z0) represents the optical path length change that is ef-fected by the sample. For areas without a sample, ∆ϕHP (x, y, z0) is estimatedby a mathematical model [31,35]:

∆ϕHP (x, y, z0) = φR(x, y, z0) − φO0(x, y, z0)= 2π

(Kxx2 + Kyy2 + Lxx + Lyy

). (9.8)

The parameters Kx,Ky in (9.8) describe the divergence of the object wave andthe properties of the applied microscopy lens. The constants Lx, Ly denote thelinear phase difference between O and R due to the off-axis geometry of theexperimental setup. For quantitative phase measurement from IHP(x, y, z0)in a first step, the complex object wave O(x, y, z = z0) in the hologramplane is determined pixel wise by solving a set of equations that is obtainedfrom insertion of (9.8) in (9.7). For that purpose, neighboring intensity val-ues within a square area of 5 × 5 pixels around a given hologram pixel areconsidered by application of a spatial phase shifting algorithm (for detailssee [9] and [23]). The utilized algorithm is based on the assumption thatonly ∆ϕHP (x, y, z0) = φR(x, y, z0) − φO0(x, y, z0) between the object waveO(x, y, z0) and the reference wave R(x, y, z0) varies rapidly spatially in thehologram plane. In addition, because of the spatial phase shifting algorithm,the object wave’s intensity has to be assumed constant within an area of about5 × 5 pixels around a given point of interest of the hologram. These require-ments can be fulfilled by an adequate relation between the magnification ofthe microscope lens and the image recording device. Therefore, the magnifi-cation of the microscope lens is chosen in such a way that the smallest imagedstructures of the sample that are restricted by the resolution of the opticalimaging system due to the Abbe criterion are over sampled by the CCD sen-sor. In this way the lateral resolution of the reconstructed holographic phasecontrast images is not decreased by the spatial phase shifting algorithm [31].

The parameters Kx,Ky, Lx, Ly in (9.8) cannot be obtained directly fromthe geometry of the experimental setup with an adequate accuracy and forthis reason are adapted once before the measurements by an iterative fittingprocess in an area of the hologram without sample [31,34].

The evaluation of digital holographic phase contrast images requires, incorrespondence to microscopy with white light illumination, a sharply focused


image of the sample. For the case that the object is not imaged sharply inthe hologram plane HP during the hologram recording process, e.g., due tomechanical instability of the experimental setup or thermal effects, in a secondevaluation step a further propagation of the object wave to the image plane canbe carried out for subsequent focus correction. The propagation of O(x, y, z0)to the image plane zIP that is located at zIP = z0 + ∆z in the distance ∆zto HP can be carried out by a Fresnel transformation [26, 31, 35] or by aconvolution algorithm [34,36,37]:

O (x, y, zIP = z0 + ∆z) = F−1{F {O (x, y, z0)} exp

(iπλ∆z

(ν2 + µ2

))}.

(9.9)

In (9.9) λ is the applied laser light wave length, ν, µ are the coordinates infrequency domain and F denotes a Fourier transformation. During the propa-gation process, the parameter ∆z is chosen in such a way that the holographicamplitude image appears sharply, in correspondence to a microscopic imageunder white light illumination. A further criterion for a sharp image of thesample is that diffraction effects due to the coherent illumination appear min-imized in the reconstructed data. As a consequence of the applied algorithmsand the parameter model for the phase difference model ∆ϕHP in (9.8), the re-sulting reconstructed holographic images do not contain the disturbing terms“twin image” and “zero order.” In addition, the method allows in compari-son to propagation by Fresnel transformation, as e.g., in [31, 35], a sharplyfocused image of the sample in the hologram plane. The propagation of O by(9.9) enables in this way the evaluation of image plane holograms containinga sharply focused image of the sample and effects no change of the image scaleduring subsequent refocusing. In the special case that the image of the sam-ple is sharply focused in the hologram plane with ∆z = 0 and thus zIP = z0,the reconstruction process can be accelerated because no propagation of O by(9.9) is required.

From O(x, y, zIP), in addition to the absolute amplitude |O(x, y, zIP)| thatrepresents the image of the sample, the phase information ∆ϕS(x, y, z0) of thesample is reconstructed simultaneously:

∆ϕS(x, y, z0) = φO (x, y, z0) − φO0(x, y, z0)

= arctan�{O (x, y, zIP)}�{O (x, y, zIP)} (mod2π). (9.10)

After removal of the 2π ambiguity by a phase unwrapping process [1], thedata obtained by (9.10) can be applied for quantitative phase contrast mi-croscopy, which is the main topic of interest for the described experiments.For an incident light geometry as depicted in the left panel of Fig. 9.13, thetopography zs can be calculated on the phase distribution ∆ϕS(x, y, z0):

zs(x, y, z0) =λ∆ϕS(x, y, z0)

2.2π=

λ

4π∆ϕs(x, y, z0). (9.11)


For illumination in transmission the thickness of cells dcell in cell culturemedium with a homogenous refractive index nmedium can be determined bymeasuring the optical path length change ∆ϕcell of the cells to the surroundingmedium:

dcell(x, y, z0) =λ∆ϕcell(x, y, z0)

2π

1ncell − nmedium

, (9.12)

with the integral refractive index ncell and the wave length λ of the appliedlaser light. For fully adherently grown cells, the parameter dcell is estimatedin first order to describe the shape of single cells [34,38].

Figure 9.15 illustrates the evaluation process of digital recorded holo-grams. Figure 9.15a,b shows a digital hologram obtained from a living humanpancreas carcinoma cell (Patu8988T) with an inverse microscope arrange-ment in transmission mode (×40 microscope lens, NA = 0.65) and the re-constructed holographic amplitude image that corresponds to a microscopicbright field image at coherent laser light illumination. Figure 9.15c depicts thesimultaneously reconstructed quantitative phase contrast image modulo 2π.The unwrapped data without 2π ambiguity, representing the optical pathlength changes that are affected by the sample in comparison to the sur-rounding medium due to the thickness and the integral refractive index, areshown in Fig. 9.15d. Figure 9.15e depicts a pseudo 3D plot of the data inFig. 9.15d. Figure 9.15f shows the cell thickness along the marked dashed linein Fig. 9.15d, which is determined by application of (9.12) with ncell = 1.38and nmedium = 1.337.

Fig. 9.15. Example for evaluation of digital holograms. (a) Digital hologram of ahuman pancreas carcinoma cell (Patu8988T); (b) Reconstructed holographic am-plitude image; (c) Quantitative phase contrast image (mod 2π); (d) Unwrappedphase distribution; (e) Pseudo 3D plot of the unwrapped phase image in gray levelrepresentation; (f) Calculated cell thickness along the dashed white line in (d)


9.5.3 Resolution and Numerical Focus

The resolution of digital holographic microscopy can be characterized by in-vestigating calibration test charts. The left panel of Fig. 9.16 shows the recon-structed amplitude (holographic image) of a negative USAF 1951 resolutiontest chart (illumination in transmission), recorded using a ×40 microscopelens (NA = 0.6). The magnified image section of group 9.5 (resolution limitof the test chart) represents a line width of 620 nm and is resolved clearly.The comparison with the Abbe criterion shows that the lateral resolutionis diffraction limited (in correspondence to bright field microscopy) and canbe increased by using microscope optics with higher numerical aperture [31].The right panel of Fig. 9.16 demonstrates the axial resolution for incidentlight illumination of a reflective metal surface (nano-structured chromium onchromium surface) recorded with a ×5 microscope lens (NA = 0.1). The de-picted elements represent a height of 30 nm and are resolved clearly in thereconstructed phase distribution. Because of the phase noise the axial resolu-tion is determined to be ≈5 nm. The digital reconstruction of different objectplanes from a single hologram enables a variable (subsequent) numerical fo-cus of digital holographic images (“digital holographic multi focus”) withoutadditional mechanical or optical components. Figure 9.17 demonstrates thedigital holographic refocus for the example of a semitransparent USAF1951test chart (upper panel) and for the holographic image of a living humanpancreas tumor cell of the type Patu8988S (lower panel).

Fig. 9.16. (Left) Reconstructed holographic amplitude of a negative USAF 1951test chart (digital hologram recorded in transmission mode), the elements of group9.6 correspond to a lateral resolution of 550 nm; (b) Reconstructed phase contrastimage of a reflective phase object (chromium on chromium sample) with steps of30 nm in axial direction (digital hologram recorded in reflection mode) [28]


Fig. 9.17. Amplitudes reconstructed from single recorded holograms in differentfocus planes (left : unfocussed reconstruction in the hologram plane; right : numer-ically focused). (a),(b) Semitransparent USAF 1951 test chart (×40 microscopeoptic, NA = 0.6) and (c),(d) Living human pancreas tumor cell Patu8988S (×100oil immersion optic, NA = 1.3) [28]

9.5.4 Digital Holographic Phase Contrast Microscopyof Living Cells

Investigations on living pancreas tumor cells [39, 40] were carried out todemonstrate the potential of digital holographic microscopy for the visual-ization of toxin-induced morphology changes. Therefore, pancreatic tumorcells were exposed at 37◦C to taxol. Digital holograms of selected cells wererecorded continuously over 4 h. Figure 9.18 shows the results. In the upperrow of Fig. 9.18, the obtained unwrapped phase distributions at t = 0, t = 78,


Fig. 9.18. Monitoring of a living PaTu8988S cell after adding a toxin (taxol) to thecell culture medium. Morphological changes such as cell rounding and finally thecell collapse are induced. Upper row : gray level coded unwrapped phase distributionat t = 0, t = 78, and t = 262 min after taxol addition. Lower row : Correspondingpseudo 3D representations of the phase data (Cooperation: Dr. Jurgen Schneken-burger, Department of Medicine B, University of Munster, Germany) [28]

Fig. 9.19. Cross sections through the measured optical path length changes cor-responding to the dashed white lines in the phase distributions of Fig. 9.18 (up-per row) [28]

and t = 262 min after taxol addition are depicted. The lower row of Fig. 9.18shows the pseudo 3D plot of the phase images. Figure 9.19 depicts cross sec-tions through the measured optical path length changes corresponding to thedashed lines in the upper row of Fig. 9.18. From Figs. 9.18 and 9.19 it isclearly visible that the toxin induces morphological changes as cell roundingand finally the cell collapses.


9.6 Discussion and Conclusions

This contribution gives an overview of proximal and distal endoscopic ESPIsystems as well as microscope (speckle) interferometric and digital holographicmicroscopy systems with regard to biomedical applications. The exemplarilyshown results obtained from the application of the proximal and distal en-doscopic ESPI systems on biological specimens demonstrate the ability ofESPI to detect displacements and differences in elasticity of biological tis-sues even underneath the visible surface. This could enhance conventionalendoscopic investigations by the substitution of the missing tactile sense by avisual information (“endoscopic taction”). In addition, for distal endoscopicESPI it has been shown that it is possible to develop compact speckle inter-ferometric arrangements with a lateral resolution in microscopic range. Thepresented (speckle) interferometric microscopy setup provides a simple inter-ferometer adjustment and a fast detection of microchanges even on livingorganisms. In connection with microscopy, digital holography allows a non-contact, fast, quantitative, minimal invasive, full field detection with high res-olution (<5 nm in axial direction) by quantitative holographic phase contrastimaging of reflective and transparent samples, with the possibility to trans-form the obtained phase data into an illustrative pseudo 3D representation.In this way, digital holographic microscopy enables topography measurementson nanostructured surfaces, quantitative on-line detection of drug effected cel-lular thickness/shape variations.

In conclusion, holographic interferometric metrology provides versatiletools for medical minimally invasive diagnostics and technical intracavity in-spection as well as for Life Cell Imaging in microscopy, Life Sciences, andBiophotonics.

Acknowledgments

Financial support by the German Ministry for Education and Research(BMBF) is gratefully acknowledged.

References

1. T. Kreis, Holographic Interferometry: Principles and Methods (Akademie Pub-lishing, Berlin, 1996)

2. V.P. Shchepinov, V.S. Pisarev, Strain and Stress Analysis by Holographic andSpeckle Interferometry (Wiley, Chichester, 1996)

3. M.A. Beek, W. Hentschel, Opt. Lasers Eng. 34, 101 (2000)4. B. Kemper, W. Avenhaus, D. Dirksen, A. Merker, G. von Bally, Appl. Opt. 39,

3899 (2000)5. W. Avenhaus, B. Kemper, G. von Bally, Gastrointest. Endosc. 54, 496 (2001)6. S. Schedin, G. Pedrini, H.J. Tiziani, Appl. Opt. 39, 2853 (2000)


7. W. Avenhaus, B. Kemper, S. Knoche, D. Domagk, C. Poremba, G. von Bally,W. Domschke, Lasers Med. Sci. 19, 223 (2005)

8. B. Kemper, D. Dirksen, J. Kandulla, G. von Bally, Opt. Commun. 194, 75(2001)

9. C. Depeursinge, in Digital Holography and Three dimensional Display: Princi-ples and Applications, ed. by T.C. Poon (Springer, New York, 2006), p. 95

10. G. von Bally, B. Kemper et al., in Biophotonics: Visions for Better Health Care,ed. by J. Popp, M. Strehle (Wiley, Berlin, 2006), p. 301

11. J.W. Goodman, in Laser Speckle and Related Phenomena, ed. by J.C. Dainty(Springer, Berlin, 1975), p. 9

12. T. Bothe, J. Burke, H. Helmers, Appl. Opt. 36, 5310 (1997)13. B. Kemper, J. Kandulla, D. Dirksen, G. von Bally, Opt. Commun. 217, 151

(2003)14. K. Creath, in Holographic Interferometry, ed. by P.K. Rastogi (Springer, Berlin,

1994), p. 10915. G. Pedrini, Y. Zou, H.J. Tiziani, Appl. Opt. 36, 786 (1997)16. T. Fricke-Begemann, J. Burke, Appl. Opt. 40, 5011 (2001)17. M. Takeda, H. Ina, S. Kobayashi, J. Opt. Soc. Am. 72, 156 (1982)18. G. von Bally, in International Trends in Applied Optics, ed. by A.H. Guenther

(SPIE Press, 2002), p. 57119. G. von Bally, Int. J. Optoelectronics 6, 491 (1998)20. S. Schedin, G. Pedrini, H.J. Tiziani, A.K. Aggarwal, Appl. Opt. 40, 2692 (2001)21. B. Kemper, S. Knoche, J. Kandulla, G. von Bally, VDI Rep. 1694, 55 (2002)22. S. Knoche, B. Kemper, J. Kandulla, G. Wernicke, G. von Bally, Opt. Commun.

270, 68 (2007)23. B. Kemper, S. Lai, A. Merker, G. von Bally, in Series on Optics Within Life

Sciences (OWLS V), ed. by C. Fotakis, T. Papazoglou, C. Kalpouzos (Springer,Heidelberg, 2000), p. 29

24. O.J. Løkberg, Opt. Lasers Eng. 26, 313 (1997)25. G. Gulker, K.D. Hinsch, A. Kraft, Opt. Lasers Eng. 36, 501 (2002)26. U. Schnars, J. Opt. Soc. Am. A 11, 2011 (1994)27. U. Schnars, W. Juptner, Meas. Sci. Technol. 13, R85 (2002)28. B. Kemper, D. Carl, A. Hoink, G. von Bally, I. Bredebusch, J. Schnekenburger,

Proc. SPIE 6191, 61910 (2006)29. T. Kreis, Handbook of Holographic Interferometry: Optical and Digital Methods

(Wiley, New york, 2005)30. P. Marquet, B. Rappaz, P.J. Magistretti, E. Cuche, Y. Emery, T. Colomb, C.

Depeursinge, Opt. Lett. 30, 468 (2005)31. D. Carl, B. Kemper, G. Wernicke, G. von Bally, Appl. Opt. 43, 6536 (2004)32. M. Liebling, T. Blu, M. Unser, J. Opt. Soc. Am. A 21, 367 (2004)33. C.J. Mann, L. Yu, C.M. Lo, M.K. Kim, Opt. Exp. 13, 8693 (2005)34. B. Kemper, D. Carl, J. Schnekenburger, I. Bredebusch, M. Schfer, W. Domschke,

G. von Bally, J. Biomed. Opt. 11, 034005 (2006)35. E. Cuche, P. Marquet, C. Depeursinge, Appl. Opt. 38, 6694 (1999)36. T.M. Kreis, M. Adams, W.P.O. Juptner, Proc. SPIE 3098, 224 (1997)37. T. Demotrakooulos, R. Mittra, Appl. Opt. 13, 665 (1974)38. B. Rappaz, P. Marquet, E. Cuche, Y. Emery, C. Depeursinge, P.J. Magistretti,

Opt. Exp. 13, 9361 (2005)


39. J. Schnekenburger, J. Mayerle, B. Krger, I. Buchwalow, F.U. Weiss, E. Albrecht,V.E. Samoilova, W. Domschke, M.M. Lerch, Gut 54, 1445 (2005)

40. H.P. Elsasser, U. Lehr, B. Agricola, H.F. Kern, Virchows Arch. B. Cell Pathol.Incl. Mol. Pathol. 61, 295 (1992)

10

Biomarkers and Luminescent Probesin Quantitative Biology

M. Zamai, G. Malengo, and V.R. Caiolfa

The genome sequence of many organisms is now complete. “However, genomesequences alone lack spatial and temporal information and are therefore asdynamic and informative as census lists or telephone directories” [1]. We knowthe words of the genetic code, yet we need to associate each of them to aspecific function. The challenge of this century is to figure out how proteinswork together to make living cells and organisms. Proteins are essential formost biological processes, but they are not simply objects with chemicallyreactive surfaces, and it is not easy to understand their function. Proteinslocalize to specific environments in the cells: membranes, cytosol, organelles,or nucleoplasm, undergo diffusive or directed movement, and often are coupledto chemical events. The exceptional capability of the proteins to regulatevirtually all dynamic processes in living cells depends on the fine regulationof their topology, movement, and chemistry. In fact, cells respond to stimulithrough signaling pathways that orchestrate the recruitment and assemblyof proteins into biomolecular machines. Roadmap1 focuses on ways to solvethe structures of protein machines. To achieve these goals, it is critical tounderstand how the cell architecture influences the formation of these proteincomplexes. Research is, therefore, turning to the study of protein function intheir most natural context [2].

10.1 Fluorophores and Genetic Dyes

10.1.1 Small Organic Dyes and Quantum Dots

Small organic fluorophores (<1 kDa) for covalent labeling of macromoleculeshave been optimized for wavelength range, brightness (extinction coefficientfor absorbance, fluorescence quantum yield) photostability, and reduction inself-quenching. Hundreds of such dyes are commercially available [3]. However,

1 (http://www.nihroadmap.nih.gov)

178 M. Zamai et al.

A B

Ccoreshellpolymer coating

biomolecule

10 nm 30 nm

Fig. 10.1. Biomarkers gallery. (a) Confocal maximal projection of fixed HEK293cells immunostained for actin (red) and stained for the nuclei with Hoechst 3342(blue). Only one cell expresses EGFP-labeled uPAR (green). (b) Confocal maximalprojection of fixed HEK293 stably transfected with EGFP-labeled uPAR (green),and stained with Cy5-uPA ligand (red). (c) (Top) Scheme of a quantum dot and(bottom) an example of application in cell imaging representing the distribution ofGM1 gangliosides found in the plasma membrane in live HeLa cells; nucleus wasstained with Hoechst 3342 (reproduced from [4])

these dyes lack specificity for any particular protein; most applications useantibodies in fixed and permeabilized cells or specifically labeled ligands(Fig. 10.1a,b).

Quantum dots (QDs) are inorganic nanocrystals that fluoresce at sharpand discrete wavelengths depending on their size [4–6]. Their extinctioncoefficients is 10–100 times higher than small fluorophores and fluorescentproteins (FPs). They also have good quantum yields and exceptional photo-stability that allow repeated imaging of single molecules [4]. Their absorbanceextends from short wavelengths up to just below the emission wavelength,so that a single excitation wavelength readily excites QDs of multiple emis-sion maxima. QDs typically contain a core of CdSe or CdTe and ZnS shell(Fig. 10.1c). For biological applications, a coating that makes QDs water sol-uble is necessary to prevent quenching by water, and allow conjugation toprotein targeting molecules such as antibodies and streptavidin. The largesize of QDs conjugated to biomolecules (10–30 nm) prevents efficient traver-sal of intact membranes, which restricts their use to permeabilized cells orextracellular or endocytosed proteins.

10.1.2 Fluorescent Proteins

Rapid advances in live-cell imaging technologies, combined with the use ofgenetically encoded fluorescent proteins (FPs), has resulted in a revolution incell biology, as it is now possible to track the assembly of protein complexeswithin the organized microenvironment of the living cell. In the next sections,we discuss some of these advances, focusing on fluorescence imaging and spec-troscopic techniques that are front-edge for the analysis of protein movementand interactions in living cells.

10 Biomarkers and Luminescent Probes in Quantitative Biology 179

Since the discovery of the GFP in 1961 by Osamu Shimomura andcolleagues [7], 30 years had to pass before the gene for GFP was cloned andthe 238 amino acids sequence determined by Prasher [8]. Douglas Prasher wasthe first person to realize the potential of GFP as a biomarker for proteins incells [9]. The manufacture of proteins using the instructions from the gene iscalled protein expression. Prasher envisioned that it would be possible to usebiomolecular techniques to insert the GFP gene at the end of the gene of anyprotein, right before the stop codon (Fig. 10.2, left). Encoded in the DNA issome type of index that directs the molecular machinery to the start of thegene of each necessary protein. When new protein is required, the gene is read,and the protein is manufactured. At the end of the gene is a message called astop codon, which ends protein production. When the cell needs to make thatprotein, it would go to the specific gene, use the information encoded in thegene to make it, but instead of stopping when the protein was made, this cellwould carry on making GFP until it reached the stop codon at the end of theGFP gene. As a result, the cell would produce a “chimeric” (e.g., modified)protein with a GFP attached to it.

A variety of techniques have been developed to construct FP fusion prod-ucts and enhance their expression in mammalian and other systems. Theprimary vehicles for introducing FP chimeric gene sequences into cells aregenetically-engineered bacterial plasmids and virus that act as vectors andtransfer the genetic information into the cell either transiently or stably(Fig. 10.2, right). In transient, or temporary, gene transfer experiments (of-ten referred to as transient transfection), plasmid or viral DNA introducedinto the host organism does not necessarily integrate into the chromosomesof the host cell, but can be expressed in the cytoplasm for a short period of

genefor protein

stop codefor protein

stop codefor protein

lipofectionreagent cell

membrane

DNA-lipidcomplex

endocytosisendosome

releasedDNA

plasmidvectorDNA

genefor protein

gene for GFP

Fig. 10.2. Left : Fusion Construct. Adapted from [10]. Right : Lipid-mediated trans-fection in mammalian cells

180 M. Zamai et al.

time. Expression of gene fusion products usually takes place over a period ofseveral hours after transfection and continues for 72–96 h after introductionof plasmid DNA into mammalian cells. In many cases, the plasmid DNA canbe incorporated into the genome in a permanent state to form stably trans-formed cell lines. The choice of transient or stable transfection depends uponthe target objectives of the investigation.

Structure of GFP

There are several reasons why GFP is a powerful biomarker. GFP is easy todetect by its fluorescence in cells; therefore, it is more versatile than mostother bioluminescent molecules that require the addition of other substancesbefore they glow. For example, firefly luciferase requires ATP, magnesium,and luciferin before it luminesces. It is a fairly small and compact protein(molecular weight, 27 kDa). The small size does not hinder the proper func-tion of the protein to which it is attached, and particularly its intracellulartrafficking and translocation. Among the most important aspects of the GFPto appreciate is that the entire 27 kDa native peptide structure is essential tothe development and maintenance of its fluorescence. It is remarkable that thefluorophore derives from a triplet of adjacent amino acids: the serine, tyrosine,and glycine residues at locations 65, 66, and 67 (referred to as Ser65, Tyr66,and Gly67; in Fig. 10.3).

Ser 65 Gly 67Tyr 66

+ 2H+

− H2O

− H+

475 nm

395 nm

Quinone

Tyr 66

PrematurationChromophore

CyclizationReaction Sites

ImidazolinoneRing System

Fig. 10.3. GFP-chromophore and the β-barrel structure. Within the hydrophobicenvironment in the center of the GFP, a reaction occurs between the carboxyl car-bon of Ser65 and the amino nitrogen of Gly67 that results in the formation of animidazolin-5-one heterocyclic nitrogen ring system. Further oxidation results in con-jugation of the imidazoline ring with Tyr66 and maturation of a fluorescent species.It is important to note that the native GFP fluorophore exists in two states. A pro-tonated form that has an excitation maximum at 395 nm, and a unprotonated onethat absorbs at approximately 475 nm. However, regardless of the excitation wave-length, fluorescence emission has a maximum peak wavelength at 507 nm, althoughthe peak is broad and not well defined


Although this simple amino acid motif is commonly found throughout innature, it does not generally result in fluorescence. What is unique to thefluorescent protein is that the location of this peptide triplet resides in thecenter of a remarkably stable barrel structure consisting of 11 β-sheets foldedinto a tube (Fig. 10.3). Since 1992, the FPs have become widely used as non-invasive markers in living cells, and their successful integration into varietyof living systems illustrates that the expression of these proteins in cells iswell tolerated. Virtually, any protein can be tagged with GFP, the resultingchimera often retains parent-protein targeting and function when expressedin cells, and therefore can be used as a fluorescent reporter to study proteindynamics.

The Future of Fluorescent Proteins

The GFP has been engineered to produce a vast number of variously coloredmutants, fusion proteins, and biosensors that are broadly referred to as FPs.The GFP sequence has been modified for optimizing the expression in differentcell types as well as the generation of GFP variants with more favorablespectral properties, including increased brightness, relative resistance to theeffects of pH variation on fluorescence, and photostability. References [1, 3]illustrate in detail the biochemical and fluorescent properties of GFP-variants.The scientist who mostly contributed to understand how GFP works anddeveloped new techniques and mutants is Roger Tsien. His group has obtainedmutants that start fluorescing faster than wild type GFP, are brighter andhave different colors (Fig. 10.4) [3].

The latest generation of jellyfish variants has solved most of the deficien-cies of the first generation FPs. The search for a monomeric, bright, andfast-maturing red FP has resulted in several new and interesting classes ofFPs, particularly those derived from coral species. Development of existingFPs, together with new technologies, such as insertion of unnatural aminoacids, will further expand the color palette. The current trend in fluorescentprobe technology is to expand the role of dyes that fluoresce into the far redand near infrared. In mammalian cells, both autofluorescence and the absorp-tion of light are greatly reduced at the red end of the spectrum. Thus, thedevelopment of far red fluorescent probes would be extremely useful for theexamination of thick specimens and entire animals. Given the success of FPsas reporters in transgenic systems (Fig. 10.5), the use of far red FPs in wholeorganisms will become increasingly important in the coming years. Finally,the tremendous potential in FP applications for the engineering of biosensorsis just now being realized. The success of these endeavors certainly suggeststhat almost any biological parameter will be measurable using the appropriateFP-based biosensor.

182 M. Zamai et al.

mP

lum

mG

rape2

mG

rape1

mC

herry

mR

aspberry

tdTom

ato

mO

range

mB

anana

mH

oneydew

YF

P (C

itrine)

EG

FP

EC

FP

EB

FP

mS

trawberry

Table 1

Class ProteinExcitation(nm)

Emission(nm) Brightness Photostability pKa Oligomerization

Properties of the best FP variants

mT

angerine

Far-red mPlum 590 649 4.1 53 <4.5 Monomer

Red mcherry 587 610 16 96 <4.5 Monomer

tdTomato 554 581 95 98 4.7 Tandem dimer

mStrawberry 574 596 26 15 <4.5 Monomer

J-Red 584 610 8.8 13 5.0 Dimer

DsRed-monomer 556 586 3.5 16 4.5 Monomer

Orange mOrange 548 562 49 9.0 6.5 Monomer

mKO 548 559 31 122 5.0 Monomer

Yellow-green mCitrine 516 529 59 49 5.7 Monomer

Venus 515 528 53 15 6.0 Weak dimer

YPetg 517 530 80 49 5.6 Weak dimer

EYFP 514 527 51 60 6.9 Weak dimer

Green Emerald 487 509 39 0.69 6.0 Weak dimer

EGFP 488 507 34 174 6.0 Weak dimer

Cyan CyPet 435 477 18 59 5.0 Weak dimer

mCFPm 433 475 13 64 4.7 Monomer

Cerulean 433 475 27 36 4.7 Weak dimer

UV-excitable green T-Sapphire 399 511 26 25 4.9 Weak dimer

Fig. 10.4. Fluorescent protein palette. Engineered FPs cover the full visible spec-trum. Top: Protein samples were purified from E. coli expression systems, excitedat wavelengths up to 560 nm and photographed by their fluorescence. Bottom: Prop-erties of the GFP variants (Adapted from [3])

Fig. 10.5. Past and present of green fluorescent protein (GFP). 1961: The jellyfishAequorea and its light-emitting organs. 1992: The amino acids sequence of GFP.2007: Mouse with brain tumor expressing the fluorescent protein variant DsRed; thezoomed area shows the tumor mass (grey) and the blood vessels expressing GFP(white) that are growing in the tumor mass


10.2 Microspectroscopy in Quantitative Biology:Where and How

Static localization of the macromolecules and proteins in cells and tissues isnot sufficient by itself to elucidate the key mechanisms that regulate funda-mental processes of the cell. The activity of a biological molecule is not justa function of its structure but also of its dynamic behavior in the cell i.e.,localization and interactions and their modifications in time. Thus, great im-portance has to be given to the spatio-temporal dynamics of the molecularinteractions within cells in situ and, in particular, in vivo. In the followingsections, we describe two spectro-microscopy approaches that, combined withthe FPs chimera technology, are of growing importance in modern biology.

10.2.1 Fluorescence Correlation Spectroscopy

Fluorescence correlation spectroscopy (FCS) was introduced by Elson, Magde,and Webb in 1972 [11]. In 1990, Denk and Webb demonstrated a new type ofmicroscope on the basis of two-photon excitation of molecules [12]. In 1995,Berland, So, and Gratton put together the two technologies, two-photon exci-tation microscopy and FCS, and demonstrated the potential of this method-ology in intracellular measurements [13]. FCS is now a widely used technique.The reader is invited to refer to the many good articles published on thephysics of the two-photon excitation process [14–16] and on the FCS tech-nique [17–20]. Here, we discuss in specific the use of FCS in cell biology asit provides information about mobility (diffusion coefficients), concentration(number of particles), association (molecular brightness), and localization (im-age) of the target molecules. All this information helps understanding thecomplex molecular interaction networks, which are at the basis of cellularprocesses.

Chemical reactions and diffusion of molecules in solutions can be followedby a variety of methods but when we need to study them in living cells, mostmethods fail abruptly. The appeal of FCS is that, especially using two-photonexcitation, living cells can be explored anywhere by a subfemtoliter volumewithout disrupting the cell integrity. The most stringent requirement for FCSto work is the possibility to observe the fluorescence signal in a small volumeand at very high sensitivity and dynamic range (Fig. 10.6).

Only if the volume is so small at any instant of time, it might contain justone or few molecules. Because of these requirements, FCS is also known asa single molecule spectroscopy. If the number of fluorescent molecules in thevolume does not change with time and if the quantum yield of the fluorophoreis constant, then the average number of the emitted photon is constant. How-ever, the instantaneous number of detected photon is not constant, becauseof the nature of the emission/detection process, which follows the Poissonstatistics (2). This added shot-noise is independent of time.2 The Poisson distribution describes events that occur rarely in a short period

184 M. Zamai et al.

2-photon volume

Delay time(s)

B C

A

Delay time(s)

counts per Bin

Aut

o C

orre

latio

n

Aut

o C

orre

latio

n

0.000.0 0.5 1.0

0.0001 0.001 0.01

0.01

0.00

0.02

0.03

0.04

0.05

0.1 1 10 100

1.5 2.0 00.0001

0.001

0.01

0.1

1

10

100

0.14000

6000

8000

10000

12000

0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1

1 2 3 4 5 6

0.01

0.02

0.03

0.04

0.05

freq

uenc

y

CP

STime (s)

PC

Ham

plitude

FCS - τ

cell

(< 10−15≈ 1µm3)

Fig. 10.6. Principle of fluorescence fluctuation spectroscopy. The number of mole-cules can change because of diffusion in and out of the volume, then fluorescenceintensity fluctuates. (a) The time of the diffusion process causes characteristic fre-quencies to appear in the fluorescence intensity trace. (b) The time structure isanalyzed by the autocorrelation function (ACF); Inset : original data are plottedin log scale. (c) The amplitude of the fluctuation depends on brightness (i.e., theaggregation state of a protein) and it can be analyzed by PCH

Instead, if the number of molecules in the excitation volume changes, evenif the quantum yield is constant, the fluorescence intensity will change withtime (Fig. 10.6a). The number of molecules can change because they can dif-fuse out of the volume. The time of the diffusion process causes characteristicfrequencies to appear in the fluorescence intensity trace. Assume that we havetwo molecules in the volume and one leaves, the relative change in intensitywill be one-half. However, if there are 10 molecules in the excitation volume

P (x, µ) =e−µµx

X!, (10.1)

where x = 0,1,2,3, (number of emitted photons) and µ = mean number of successesin the given time interval or region of space (number of counts). If µ is the averagenumber of successes occurring in a given time interval or region in the Poissondistribution, then the mean and the variance of the Poisson distribution are bothequal to µ: E(x) = m, and V (x) = σ e2 = µ. In a Poisson distribution, only oneparameter, µ, is needed.


and one of them leaves, the relative change will be only 1/10. This meansthat the smaller the ratio fluctuation/average-signal, the larger the number ofmolecules in the volume. It can be shown that this ratio is exactly proportionalto the inverse of the number of molecules in the volume of excitation [21]. Thisrelationship allows the measurements of the number of molecules in a givenvolume in the interior of cells [13, 22]. Molecular heterogeneity in the cell ormolecular reactions can also induce the simultaneous presence of molecules ofdifferent kind in the same volume [23]. One important case is when proteinsbind each other forming a molecular aggregate. Let us consider two identi-cal proteins with one fluorescent probe each forming a molecular dimer. Thedimer is different than the monomer because it carries twice the number of flu-orescent moieties (Fig. 10.6). When one of these aggregates enters the volumeof excitation, it will cause a larger fluctuation of the intensity than a singlemonomer. Clearly, the amplitude of the fluctuation carries information on thebrightness of the diffusing particle (i.e., monomer vs. dimer). Thus, both timeand amplitude structures are affected by underlying molecular species, andthe dynamic processes that cause the change of the fluorescence intensity.The statistical analysis of the fluctuations of the fluorescence signal has torecover both information (Fig. 10.6a).

Analysis of the Time Structure of the Fluctuating Signal

The analysis of the time structure of the fluctuating signal is typically doneby autocorrelation analysis. The autocorrelation function, ACF = G(τ), char-acterizes the time-dependent decay of the fluorescence fluctuations to theirequilibrium value (Fig. 10.6b). In simple terms, ACF calculates the similaritybetween a signal I(t), and a copy of the same signal shifted by a time lag τ ,I(t + τ):

G(τ) =〈I(t)I(t + τ)〉 − 〈I(t)〉2

〈I(t)〉2 . (10.2)

The autocorrelation function yields two parameters: the diffusion coef-ficient (D) and the average number of particles in the observation volume< N > given by the inverse of G(0), multiplied by a constant that depends onthe illumination profile. In the case of identical molecules undergoing randomdiffusion in a Gaussian illuminated volume, the characteristic autocorrelationfunction is given by the following expression [22]:

G(τ) =γ

N

(1 +

8Dτ

ω2r

)−1 (1 +

8Dτ

ω2a

)−1/2

, (10.3)

where D is the diffusion coefficient; ωr and ωa are the beam waist in the radialand in the axial directions; N is the number of molecules; γ is the numericalfactor that accounts for the nonuniform illumination of the volume; and τ isthe delay time. Other formulas have been derived for the Gaussian-Lorentzianillumination profile [13] and for molecules diffusing on a membrane [21].

186 M. Zamai et al.

Analysis of the Amplitude Structure of the Fluctuating Signal

The expression for the statistics of the amplitude fluctuations is generallygiven under the form of the histogram of the photon counts for a given sam-pling time ∆t. This is known as the photon counting histogram (PCH) dis-tribution. The PCH analysis calculates the probability of detecting photonsper sampling time [24, 25]. The probability is experimentally determined bythe histogram of the detected photons. In principle, the Poisson distributiondescribes the occupation number of particles that can freely go in and out asmall excitation volume. However, mainly due to the diffusion of molecules inthe inhomogeneous excitation volume, the distribution of photon counts devi-ates from the Poisson distribution. The analysis of the distribution of photonsis based on the deviation of the measured PCH from the expected Poissondistribution because of the molecule occupation number (Fig. 10.7). Two pa-rameters characterize the photon distribution: the number of molecules in theobservation volume (N) and the molecular brightness (ε), which is defined asthe average number of detected photons per molecule per second. The ana-lytical expression for the PCH distribution for a single molecular species of agiven brightness has been derived for the 3D-Gaussian illumination profile [24]and is reported below:

P3DG (k;V0; ε) =1V0

πω20z0

2k!

∞∫0

γ(k, εe−4x2

)dx, (10.4)

where V0 is the volume of illumination; ε is the brightness of the molecules(counts per seconds per molecule); and k is the number of photons in a givetime interval. The integral that contains the incomplete gamma function (γ)can be numerically evaluated. Similar expressions have been derived for othershapes of the illumination volume [24].

PCH is extremely useful for analysis fluctuations inside cells, where thetarget protein can associate in clusters or bind to some other preexisting ag-gregates. Diffusion coefficients for proteins in solution depend on their mole-cular weight (τdiff ∝ D−1 ∝ M1/3). However, in the case of monomer–dimerspecies, the difference between the diffusion coefficients of the two species is

position 1

Position (in time)

Inte

nsity

position 3

position 2

photons counts/ time0

10

100

1 2 3 4 5 6 7 8 9

super-poisson

poisson

Fre

quen

cy o

f eve

ntsPosition 2

Position 1Position 3

Fig. 10.7. Super-Poissonian distribution of detected photons from diffusing fluo-rophores due to (1) fluctuation of the particles number (Poissonian), (2) no uniformvolume of the PSF, and (3) photon detection statistics (Poissonian)


just 1.2. It can be shown that a mass ratio of at least 5 is necessary to detectsignificant differences in the diffusion coefficient. Instead the brightness canchange by a factor of 2 when the equilibrium moves from all monomers to alldimers (Fig. 10.8).

The all idea of FCS-PCH is the detection of fluorescence intensity fluctua-tions that can be achieved reducing the observation volume to subfemtolitersbut also having few molecules in it. In practice this means that we need con-centrations of the fluorophore in the nanomolar range. The consequences aretwofold: (1) To study binding events, the dissociation constant of the com-

Fig. 10.8. Study of molecular association and diffusion at the cell membrane ofthe ∆D1 form of the GPI-anchored uPAR, fused to EGFP and transfected in hu-man kidney cells, HEK293. Top panels: Two-photon images of live cells showing thedistribution of the receptor at the ventral, mid, and topside of the cell membrane;fluorescence in pseudo-color scale from low (blue) to high (red) intensity; (+): re-gions at which fluorescence fluctuation was recorded. Mid panel : A representativefluorescence fluctuation trace acquired in position (a). Bottom panels: Representa-tive ACF and PCH for two fluorescence fluctuation traces acquired in regions (a)and (b). PCH indicates that the mutant receptor associates in oligomers (brightnessb = 3, 900 cpsm, brightness a = 9, 100 cpsm). The ACFs of the two forms, however,do not reveal detectable differences in diffusion coefficient. Control experiments insolution have shown that the EGFP variant, lacking the receptor uPAR domains,does not dimerise at the concentrations found in the cells

188 M. Zamai et al.

plexes must be in the nanomolar range or below; (2) The expression in vivoof the FP-proteins systems must be low. This is not a trivial issue as manyproteins are segregated in intracellular and organelle compartments often atlocal concentration higher than nanomolar. FCS-PCH criteria are oppositeto those of classical confocal or multiphoton fluorescence microscopy and asystematic optimization of the expression levels of the FP-tagged proteins isnecessary.

10.2.2 Fluorescence Lifetime Imaging (FLIM)

The development of confocal and multiphoton fluorescence microscopy hasintroduced enormous prospectives in biomedical sciences. These microscopiespermit a variety of high-contrast and multidimensional imaging. However, thefluorescence of organic molecules is not only characterized by the emissionspectrum, but also possesses a distinctive lifetime. The fluorescence lifetime isdefined as the mean amount of time a fluorophore spends in the excited stateafter the absorption of an excitation photon. Lifetime can be measured bythe time domain or by the frequency domain techniques. In the time domain(Fig. 10.9a–b), a short pulse of light excites the sample, and the subsequentfluorescence emission is recorded as a function of time. This usually occurs onthe nanosecond timescale. In the frequency domain (Fig. 10.9c–d), the sam-ple is excited by a modulated source of light. The fluorescence emitted bythe sample has a similar waveform, but is modulated and phase-shifted fromthe excitation curve. Both modulation (M) and phase-shift φ are determinedby the lifetime of the sample emission; that lifetime can be calculated fromthe observed demodulation ratio and phase-shift. Both of these domains yieldequivalent data, and take advantage of the fluorescence decay law, which isbased on the first-order kinetics. The decay law postulates that if a populationof molecules is instantaneously excited when photons are absorbed, then theexcited population, and hence the fluorescence intensity as a function of time,I(t), gradually decays to the ground state. Decay kinetics can be describedby: I(t) = αe

−tτf ; where α is the intensity at time t = 0, t is the time after the

absorption, and τf is the lifetime, that is, when the fraction of the populationof molecules in the excited state (and the fluorescence intensity) has decreasedby a factor of 1/e. Note that before absorption, I(t) = 0.

Time Domain

Time-domain measurements are based on the assumption that, when photonsare absorbed, the molecules can be excited in an infinitely brief moment. Thisidea is commonly known as the delta or delta-pulse. The delta-pulse idea isused to interpret data obtained with real-pulsed light sources with measur-able pulse-widths. In practice, the time-dependent profile of the light-pulseis reconvolved with the decay-law function. Reconvolution assumes that thedelta-pulses are continuous functions, so that the observed decay is the con-volution integral of the decays from all delta-pulses initiated during the finite


Fig. 10.9. Lifetime measurements in time and frequency domains. (a) Actualpulsed light-source (gray) and sample response (black), showing the gradual de-cay of fluorescence intensity with time. In this example, the decay could be fit-ted with a mono-exponential curve (dotted) giving a lifetime of 1.309 0.003 ns.(b) TCSPC-fluorometer: A pulsed light source excites the sample repetitively. Thesample emission is observed by a detector, while the excitation flashes are detectedby a synchronization module (SYNC). A constant fraction discriminator (CFD)responds to only the first photon detected (small arrows), independent of its ampli-tude, from the detector. This first photon from sample emission is the stop signalfor the time-to-amplitude converter (TAC). The excitation pulses trigger the startsignals. The multichannel analyzer (MCA) records repetitive start–stop signals ofthe single-photon events from the TAC, to generate a histogram of photon countsas a function of time channel units. The lifetime is calculated from this histogram.(c) Excitation (black) and sample response (gray), illustrating the phase-angle shift(φ) and demodulation ratio (M). (d) Multifrequency cross-correlation fluorometer.An unmodulated light source emits a spectrum of continuous-wave light. The excita-tion monochromator (Excit. Mono.) selects an excitation wavelength. An amplified(Amp 1) master synthesizer (Master) drives the Pockels cell (Pockels) at a basefrequency, Rf , which modulates the excitation beam. The modulated beam excitesthe Sample, causing the sample to emit modulated fluorescence also at the base Rf .An emission monochromator (Emis. Mono.) selects one wavelength of modulatedfluorescence. The photomultiplier tube (PMT) is modulated by an amplified (Amp2) slave synthesizer (Slave) at the base Rf plus a low-frequency cross-correlationnote (f). The sample emission at Rf cancels the slave Rf + f frequencies to yieldthe f signal containing the same phase-angle shift (φ) and demodulation ratio (M)as the Rf fluorescence

190 M. Zamai et al.

pulse-width. The method of time-correlated single-photon counting (TCSPC)is, by far, a superior method for measuring time-domain decays (Fig. 10.9b).TCSPC uses a pulsed light-source and a circuit to detect single-photon eventsat a detector. In a repetitive series of many start–stop signals from the cir-cuitry, a binned histogram in time channels of single-photon counts is gradu-ally generated. TCSPC relies on a principle of Poissonian statistics that onlyone photon can be counted at a time and in any one channel, to avoid skewingthe time-dependent statistics in photon-pile-up. Pile-up thus limits the data-acquisition rate of TCSPC to a few (typically 12) percent of the repetitionrate. In practice, the single-photon limit is not a major hindrance becausethe pile-up limit can be monitored during the experiment, and decay timeswith sufficient photon counts in can be obtained in seconds to minutes withrepetition rates in the megahertz range. In addition, the Poissonian nature ofthe statistics allows the data to be rigorously analyzed.

Frequency Domain

The fluorescence decay parameters in the decay law impulse function may beobtained on the basis of the relation of a sinusoidally modulated excitationbeam to the fluorescence emission response (Fig. 10.9c). The emission occursat the same frequency as the excitation. Because of the loss of electron en-ergy (Stokes shift) between excitation and emission, the emission waveformis demodulated and phase-shifted in comparison to the excitation. Thus thedemodulation ratio (M) and phase-angle shift (φ) constitute two separate ob-servable parameters that are both directly related, via a Fourier transforma-tion, to the initial fluorescence intensity, α, and lifetime, τ for a populationof fluorophores. Frequency-domain measurements are best performed usingmulti-frequency cross-correlation phase-and-modulation (MFCC), shown inFig. 10.9d. A modulated beam excites the sample. The fluorescence emissionis detected by a photomultiplier tube (PMT) modulated at the same baseradio-frequency as the master plus a low cross-correlation frequency (a fewhertz). The base-frequency signals are filtered to reveal the cross-correlationfrequency signal, which contains all the same demodulation (M) and phaseangle shift (φ) information as the fluorescence emission.

Fluorescence Lifetime Imaging Microscopy (FLIM) and Forster ResonanceEnergy Transfer (FRET)

Fluorescence lifetime imaging microscopy (FLIM) is a technique to map thespatial distribution of nanosecond excited state lifetimes within microscopicimages. FLIM has proven to be a robust and established technique in moderncell biology for lifetime contrast in ion-imaging, quantitative imaging, tissuecharacterization, and medical applications [26–28], and it is especially suitablefor in situ protein–protein interaction studies using Forster Resonance EnergyTransfer (FRET) [29–31]. FRET is a photophysical phenomenon in whichenergy is transferred from the first excited electronic state (S1) of a fluorophore


(called donor D) to another nearby absorbing (but not necessarily emitting)molecule (called acceptor A). Thus, there is a concerted quenching of D andactivation of A fluorescence (Fig. 10.10). For this reason, the acronym FRETis often used to designate fluorescence resonance energy transfer. The processinvolves the resonant coupling of emission and absorption dipoles and is thusnonradiative. That is, it competes with other radiative (fluorescence) andnonradiative pathways for deactivation, resulting in a decrease of the donorlifetime. The energy transfer rate from the donor to the acceptor decreaseswith the sixth power of the distance and thus is apparent only at distancesshorter than 10 nm [32]. At the critical distance where 50% of the donorenergy is transferred to an acceptor, the Forster radius, the donor emissionand fluorescent lifetime are each reduced by 50%, and sensitized emission(acceptor emission specifically under donor excitation) is increased.

Because of its utility in reporting nanometer-scale interactions, FRET hasbecome an important tool in cell biology [33–38]. FRET in cell biology is usedcommonly to verify whether labeled proteins physically interact: by measur-ing the FRET efficiency, distances on the nanometer scale (a scale within theglobular radii of proteins) can thus be estimated using a light microscope. Bymeasuring these effects, FRET microscopic imaging can verify close molecularassociations between colocalized donor and acceptor-labeled fusion proteinsthat are far beyond the resolution of fluorescent microscopy. Obvious difficul-ties in intensity-based FRET measurements in cells is that the concentrationsof the donor and acceptor are variable and unknown, the emission band onthe donor extends into the absorption and emission band of the acceptor,and the absorption band of the acceptor commonly extends into the absorp-tion band of the donor. A further complication is that only a fraction of the

Fig. 10.10. The principle of FRET and dependence of FRET efficiency on donor–acceptor distance

192 M. Zamai et al.

donor molecules interacts with an acceptor molecule. These effects are hardto distinguish in intensity-based FRET measurements [39, 40]. In contrast,FLIM-based FRET techniques have the benefit that the results are obtainedfrom a single-lifetime measurement of the donor. These approaches do notneed calibration in different cells, and are nondestructive. The important dif-ference between TCSPC and frequency domain FLIM resides in the sourceof excitation. In frequency domain FLIM (Fig. 10.9c) as the sample is ex-cited by a sinusoidally modulated light, only single photon excitation can beapplied, which in living cells induced major photobleaching effects during life-time scans. Alternatively TCSPC-FLIM takes advantage of the two-photonlaser subnanosecond pulses (Fig. 10.9a).

However, the principal strength of TCSPC−FLIM is the statistical accu-racy, and this is reflected in the ability to fit two or sometimes three expo-nential functions to a particular fluorescence decay. The advantage is that itis often possible (depending on the quality of the data acquired) to determinethe ratio of FRET vs. non-FRET lifetimes contained within a single pixel ofan image, and therefore to generate a map of interacting vs. noninteractingproteins in the cell (Fig. 10.11). These results may reflect the relative boundand unbound fractions within a particular protein–protein interaction reac-tion. It is difficult or impossible to obtain these results directly using any otherFRET or FLIM approach. Importantly for biochemists, these parameters aredirectly available using TCSPC−FLIM, and if the concentrations of the in-teracting proteins can be estimated (this is no mean feat inside cells, but itcan be done), then estimates of association and dissociation constants can be

pulse donor

EGFPemission

Wavelength nm450

0.0

0.2

0.4

0.6

0.8

1.0

500 550 600

Inte

nsity

(ar

b. u

nits

)

mRFP1excitation

donor + acceptortD+A

FRETeff = 1_τD+AτD

τD

Fig. 10.11. FRET measured by TCSFC−FLIM. Representative fluorescencedecays are shown for cells transfected with the donor (GPI−uPAR−EGFP)and cell cotransfected with donor and acceptor receptors (GPI−uPAR−EGFP,GPI−uPAR−mRGFP1). FRET efficiency is derived from the ratio betweenquenched and unqueched lifetimes


made. The resolution of a biexponential FRET system relies in part on thecorrect selection of fluorophore, particularly for the donor.

Many FPs have been shown to have complex fluorescence decays, oftenbiexponential or even more complex, presumably due to protonation or someinternal relaxation process, even in a non-FRET system. An example of thisis ECFP (enhanced cyan fluorescent protein). This is commonly used as aFRET donor, because of its high quantum yield, long emission tail, and rel-atively blue emission. However, several fluorescence lifetime studies have re-vealed ECFP to have a biexponential lifetime [41, 42], making it difficult toresolve the interacting fraction from a noninteracting fraction in a FRET sys-tem (using FLIM); to do this, four lifetimes would need to be fitted to thedata (i.e., two for the non-FRET system, and two for the FRET system), andit is beyond most FLIM measurements to acquire accurate enough data ina practical time period. However, there are FPs useful as FRET donors inFLIM measurements. EGFP (enhanced GFP) has a high quantum yield anda monoexponential decay in a non-FRET system, and has been shown to bea good donor for red FPs, such as dsRed and mRFP [monomeric RFP (redfluorescent protein)] [43] (Fig. 10.12). One advantage of EGFP for many cell

top membrane ventral membrane Autocorrelation functions

g(ta

u) n

orm

topmembrane

tau (s)0.0001

0.00

0.25

0.50

0.75

1.00

0.001 0.01 0.1 1 10

ventralmembrane

Fluorescence Intensity imageshigh

low

cps

TCSPC-FLIM imagesFRET

0 +/− 8%

FRET18 +/− 8%

Fig. 10.12. Study of molecular association and diffusion at the cell membrane ofthe active GPI-anchored uPAR, fused to EGFP and transfected in human kidneycells, HEK293. Top panels: two-photon images of live cells showing the distributionof the receptor at the ventral and topside of the cell membrane; RepresentativeACF acquired in two regions at the top and ventral membrane (+) showing that thereceptor has longer diffusion time at the cell adhesion site (about 0.9 and 0.1 µm2/srespectively). Lower panels: FLIM analysis of cells cotransfected with the GFP-mRFP1 receptor pair showing that the regions where the receptor is slow diffusingare also regions of high FRET efficiency

194 M. Zamai et al.

biologists is that it has been in use for many years, so cDNA constructs areoften already made. More recently, the drive for a mono-exponentially decay-ing fluorescent protein led to the development of Cerulean, a variant of ECFPwith all the advantages of a cyan donor, but a monoexponential fluorescentdecay [44]. FLIM is still an emerging technology, of great interest to the cellbiologist. One reason for lack of use of FLIM in biology laboratories is thetechnical difficulty associated with making reliable measurements. Althoughthere are a number of different approaches to FLIM, some more turn-key thanothers, TCSPC can resolve additional parameters that make the technical dif-ficulties worthwhile. In particular, the ability to make statistically accurateestimates of the bound vs. unbound fractions of fusion proteins inside livingcells ought to be a very attractive prospect for cell biologists and biochemists.

10.2.3 Glossary of Molecular Biology

Vector : The DNA “vehicle” used to carry experimental DNA and to cloneit. The vector provides all sequences essential for replicating the test DNA.Typical vectors include plasmids, cosmids, phages, and YACs.

Plasmid : A circular piece of DNA present in bacteria or isolated from bac-teria. Escherichia coli, the usual bacteria in molecular genetics experiments,has a large circular genome, but it will also replicate smaller circular DNAsas long as they have an “origin of replication”. Plasmids may also have otherDNA inserted by the investigator. A bacterium carrying a plasmid and repli-cating a million-fold will produce a million identical copies of that plasmid.Common plasmids are pBR322, pGEM, pUC18.

Transfection: A method by which experimental DNA may be put intoa cultured mammalian cell. Such experiments are usually performed usingcloned DNA containing coding sequences and control regions (promoters, etc.)to test whether the DNA will be expressed. As the cloned DNA may have beenextensively modified (for example, protein-binding sites on the promoter mayhave been altered or removed), this procedure is often used to test whether aparticular modification affects the function of a gene.

Transient Transfection: When DNA is transfected into cultured cells, it isable to stay in those cells for about 2–3 days, but then will be lost (unlesssteps are taken to ensure that it is retained – see stable transfection). Duringthose 2–3 days, the DNA is functional, and any functional genes it containswill be expressed. Investigators take advantage of this transient expressionperiod to test gene function.

Stable Transfection: A form of transfection experiment designed to producepermanent lines of cultured cells with a new gene inserted into their genome.Usually, this is done by linking the desired gene with a “selectable” gene, i.e.,a gene that confers resistance to a toxin (like G418, aka Geneticin). Uponputting the toxin into the culture medium, only those cells that incorporatethe resistance gene will survive, and essentially all of those will also haveincorporated the experimenter’s gene.


Mutant Type Protein: Mutants are proteins lacking one or more biologicalactivities of the wild type parent protein.

References

1. R.Y. Tsien, Building and breeding molecules to spy on cells and tumors. FEBSLett. 579(4), 927–932 (2005)

2. R.N. Day, F. Schaufele, Imaging molecular interactions in living cells. Mol.Endocrinol. 19(7), 1675–1686 (2005)

3. N.C. Shaner, P.A. Steinbach, R.Y. Tsien, A guide to choosing fluorescentproteins. Nat. Meth. 2(12), 905–909 (2005)

4. X. Michalet, F.F. Pinaud, L.A. Bentolila, J.M. Tsay, S. Doose, J.J. Li,G. Sundaresan, A.M. Wu, S.S. Gambhir, S. Weiss, Quantum dots for live cells,in vivo imaging, and diagnostics. Science 307(5709), 538–544 (2005)

5. F. Pinaud, X. Michalet, L.A. Bentolila, J.M. Tsay, S. Doose, J.J. Li, G. Iyer,S. Weiss, Advances in fluorescence imaging with quantum dot bio-probes.Biomaterials 27(9), 1679–1687 (2006)

6. J.K. Jaiswal, S.M. Simon, Potentials and pitfalls of fluorescent quantum dotsfor biological imaging. Trends Cell Biol. 14(9), 497–504 (2004)

7. O. Shimomura, F.H. Johnson, Y. Saiga, Extraction, purification and proper-ties of aequorin, a bioluminescent protein from the luminous hydromedusan,aequorea. J. Cell Comp. Physiol. 59, 223–239 (1962)

8. D.C. Prasher, V.K. Eckenrode, W.W. Ward, F.G. Prendergast, M.J. Cormier,Primary structure of the aequorea victoria green-fluorescent protein. Gene111(2), 229–233 (1992)

9. M. Chalfie, Y. Tu, G. Euskirchen, W.W. Ward, D.C. Prasher, Green fluorescentprotein as a marker for gene expression. Science 263(5148), 802–805 (1994)

10. Marc Zimmer, Glowing Genes: A Revolution in Biotechnology (PrometheusBooks, New York, 2005)

11. D. Magde, E. Elson, W.W. Webb, Thermodynamic fluctuations in a reactingsystem: measurement by fluorescence correlation spectroscopy. Phys. Rev. Lett.29, 705–708 (1972)

12. W. Denk, J.H. Strickler, W.W. Webb, Two-photon laser scanning fluorescencemicroscopy. Science 248(4951), 73–76 (1990)

13. K.M. Berland, P.T. So, E. Gratton Two-photon fluorescence correlation spec-troscopy: Method and application to the intracellular environment. Biophys.J. 68(2), 694–701 (1995)

14. G.J. Brakenhoff, M. Muller, R.I. Ghauharali, Analysis of efficiency of two-photon versus single-photon absorption of fluorescence generation in biologicalobjects. J. Microsc. 183(Pt 2), 140–144 (1996)

15. W. Denk, K. Svoboda, Photon upmanship: Why multiphoton imaging is morethan a gimmick. Neuron 18(3), 351–357 (1997)

16. B.R. Masters, P.T. So, K.H. Kim, C. Buehler, E. Gratton, Multiphoton ex-citation microscopy, confocal microscopy, and spectroscopy of living cells andtissues; functional metabolic imaging of human skin in vivo. Meth. Enzymol.307, 513–536 (1999)

196 M. Zamai et al.

17. S. Breusegem, N. Barry, Q. Ruan, E. Ruan Gratton, J. Eid, in Fluctuation cor-relation spectroscopy in cells: Determination of molecular aggregation, ed. byKluwer and Plenum Publishers. Biophotonics-Optical Science and Engineeringfor the 21st Century (Kluwer, Dordecht, 2004), pp. 1–16

18. S.T. Hess, W.W. Webb, Focal volume optics and experimental artifacts inconfocal fluorescence correlation spectroscopy. Biophys. J. 83(4), 2300–2317(2002)

19. Y. Chen, J.D. Muller, K.M. Berland, E. Gratton, Fluorescence fluctuationspectroscopy. Methods 19(2), 234–252 (1999)

20. N.L. Thompson, A.M. Lieto, N.W. Allen, Recent advances in fluorescencecorrelation spectroscopy. Curr. Opin. Struct. Biol. 12(5), 634–641 (2002)

21. N.L. Thompson, in Fluorescence Correlation Spectroscopy, ed. by J.R.Lakowicz. Topics in Fluorescence Spectroscopy (Plenum, New York, 1991),pp. 337–378

22. H. Qian, E.L. Elson, Distribution of molecular aggregation by analysis offluctuation moments. Proc. Natl. Acad. Sci. USA 87(14), 5479–5483 (1990)

23. A.G. Palmer III, N.L. Thompson, Molecular aggregation characterized byhigh order autocorrelation in fluorescence correlation spectroscopy. Biophys.J. 52(2), 257–270 (1987)

24. Y. Chen, J.D. Muller, P.T. So, E. Gratton, The photon counting histogram influorescence fluctuation spectroscopy. Biophys. J. 77(1), 553–567 (1999)

25. J.D. Muller, Y. Chen, E. Gratton, Resolving heterogeneity on the single mole-cular level with the photon-counting histogram. Biophys. J. 78(1), 474–486(2000)

26. E.B. van Munster, T.W. Gadella, Fluorescence lifetime imaging microscopy(FLIM). Adv. Biochem. Eng. Biotechnol. 95, 143–175 (2005)

27. D. Elson, J. Requejo-Isidro, I. Munro, F. Reavell, J. Siegel, K. Suhling,P. Tadrous, R. Benninger, P. Lanigan, J. McGinty, C. Talbot, B. Treanor,S. Webb, A. Sandison, A. Wallace, D. Davis, J. Lever, M. Neil, D. Phillips,G. Stamp, P. French, Time-domain fluorescence lifetime imaging applied tobiological tissue. Photochem. Photobiol. Sci. 3(8), 795–801 (2004)

28. I. Munro, J. McGinty, N. Galletly, J. Requejo-Isidro, P.M. Lanigan, D.S. Elson,C. Dunsby, M.A. Neil, M.J. Lever, G.W. Stamp, P.M. French, Toward theclinical application of time-domain fluorescence lifetime imaging. J. Biomed.Opt. 10(5), 051403 (2005)

29. Y. Chen, J.D. Mills, A. Periasamy, Protein localization in living cells andtissues using FRET and FLIM. Differentiation 71(9–10), 528–541 (2003)

30. Y. Chen, A. Periasamy, Characterization of two-photon excitation fluorescencelifetime imaging microscopy for protein localization. Microsc. Res. Tech. 63(1),72–80 (2004)

31. H. Wallrabe, A. Periasamy, Imaging protein molecules using FRET and FLIMmicroscopy. Curr. Opin. Biotechnol. 16(1), 19–27 (2005)

32. L. Stryer, Fluorescence energy transfer as a spectroscopic ruler. Annu. Rev.Biochem. 47, 819–846 (1978)

33. T.C. Voss, I.A. Demarco, R.N. Day, Quantitative imaging of protein interac-tions in the cell nucleus. Biotechniques 38(3), 413–424 (2005)

34. R.N. Day, D.W. Piston, Spying on the hidden lives of proteins. Nat. Biotech-nol. 17(5), 425–426 (1999)


35. P.I. Bastiaens, A. Squire, Fluorescence lifetime imaging microscopy: spatialresolution of biochemical processes in the cell. Trends Cell Biol. 9(2), 48–52(1999)

36. R. Pepperkok, A. Squire, S. Geley, P.I. Bastiaens, Simultaneous detection ofmultiple green fluorescent proteins in live cells by fluorescence lifetime imagingmicroscopy. Curr. Biol. 9(5), 269–272 (1999)

37. A. Squire, P.I. Bastiaens, Three dimensional image restoration in fluorescencelifetime imaging microscopy. J. Microsc. 193(Pt 1), 36–49 (1999)

38. A. Periasamy, R.N. Day, Visualizing protein interactions in living cells usingdigitized GFP imaging and FRET microscopy. Meth. Cell Biol. 58, 293–314(1999)

39. Y. Gu, W.L. Di, D.P. Kelsell, D. Zicha, Quantitative fluorescence resonance en-ergy transfer (FRET) measurement with acceptor photobleaching and spectralunmixing. J. Microsc. 215(Pt 2), 162–173 (2004)

40. R.R. Duncan, Fluorescence lifetime imaging microscopy (FLIM) to quan-tify protein-protein interactions inside cells. Biochem. Soc. Trans. 34(Pt 5),679–682 (2006)

41. R.R. Duncan, A. Bergmann, M.A. Cousin, D.K. Apps, M.J. Shipston, Multi-dimensional time-correlated single photon counting (TCSPC) fluorescence life-time imaging microscopy (FLIM) to detect fret in cells. J. Microsc. 215(Pt 1),1–12 (2004)

42. L.M. Felber, S.M. Cloutier, C. Kundig, T. Kishi, V. Brossard, P. Jichlinski,H.J. Leisinger, D. Deperthes, Evaluation of the CFP-substrate-YFP systemfor protease studies: advantages and limitations. Biotechniques 36(5), 878–885(2004)

43. M. Peter, S.M. Ameer-Beg, M.K. Hughes, M.D. Keppler, S. Prag, M. Marsh,B. Vojnovic, T. Ng, Multiphoton-film quantification of the EGFP-mRFP1FRET pair for localization of membrane receptor-kinase interactions. Biophys.J. 88(2), 1224–1237 (2005)

44. M.A. Rizzo, G.H. Springer, B. Granada, D.W. Piston, An improved cyan flu-orescent protein variant useful for FRET. Nat. Biotechnol. 22(4), 445–449(2004)

11

Fluorescence-Based Optical Biosensors

F.S. Ligler

11.1 Introduction

Biosensors integrate biological molecules with a signal transduction deviceto produce a signal when a molecular recognition event occurs. This tutorialwill focus entirely on biosensors that employ optical devices and incorporatefluorescent mechanisms for signal transduction. Compared with the electro-chemical glucose sensors found in pharmacies world wide, optical biosensorsare often more complex and more costly. However, optical biosensors are bet-ter suited for repetitive analysis or continuous monitoring, for interrogation ofcomplex fluids, and for measuring binding events in real time. Optical imagingmethods have also been widely adapted to measuring microarrays of recog-nition events; this experience provides a base for the development of highlymultiplexed optical biosensors.

Optical biosensors not included in the following discussion are, nonetheless,worth mentioning. This group of biosensors is primarily directed at detectionof a target without the requirement for a label. Thus performing the assayis simplified by requiring only the exposure of the sample (containing target)to the biological recognition molecule. Noteworthy among these methods areinterferometry, surface plasmon resonance, resonant and antiresonance reflec-tometry, and cantilever-based systems [1, 2]. All of these sensors measure achange in optical properties, usually refractive index, at the sensing surfacewhere the recognition molecules are immobilized. Convenient to use, theyare excellent tools for measuring reactions in well-defined fluids. They tendto be less sensitive to very small targets or to very large targets that havemost of their mass outside the sensing region; improvements in waveguidetechnology are minimizing these problems. However, nonspecific adsorptionof components from complex samples can reduce the sensitivity by creating asignificant background signal that must be accurately subtracted.

The following tutorial on fluorescence-based optical biosensors is orga-nized into five parts: (1) biological recognition molecules and assay formats,(2) displacement immunosensors, (3) fiber optic biosensors, (4) bead-based

200 F.S. Ligler

biosensors, and (5) planar biosensors. These five groups are certainly not anexhaustive coverage of fluorescence-based biosensors, but should provide thereader with at least a general appreciation of the breadth of options available.All of these sensors include the same basic components: biological recognitionmolecules, a source of excitation light, and an optical readout device capa-ble of discriminating excitation light from emitted fluorescence. Most of thebiosensors also include a fluid transfer system to move samples and reagentsover the sensing surface. The components and configurations for these systemsdepend on the geometry of the sensing surface and the degree of automationof the biosensor.

11.2 Biological Recognition Moleculesand Assay Formats

Optical biosensors have utilized a wide variety of biological recognition mole-cules. With the exception of a few instances where enzyme catalysis is mea-sured, they employ affinity or binding functions. The most often employedmolecules are antibodies and oligonucleotides because they are readily avail-able and the binding reactions have undergone extensive analysis at the mole-cular level. However, oligosaccharides, antibiotic peptides, siderophores, andcombinatorial molecules have also been utilized.

No matter what the binding molecule is, the method by which it is im-mobilized on the sensing surface is very important if that binding function isto be preserved. In general, the surface is first modified with a silane or thiolto provide reactive groups for subsequent attachment of the biomolecules; ifpossible, this layer also helps prevent subsequent nonspecific binding of thebiorecognition molecule that would result in denaturation. Next a crosslinkeris attached to the free end of the silane or thiol and binds the biorecogni-tion molecule [3]. For antibody immobilization, many studies have focused onorienting the antibody to avoid interference with the binding site. Althoughattachment through thiol groups in antibody fragments or the carbohydrateon the region of the molecule distal to the active sites achieves this goal,preventing secondary nonspecific adsorption is often even more critical.

In a popular variation of the silane-based immobilization approach, avidinis immobilized instead of the biorecognition molecule, and the biorecognitionmolecule is biotinylated and bound to the avidin through a high affinity, non-covalent bridge. When plastic is used instead of glass or silicon, nonspecificadsorption of the avidin provides a fully functional surface [4, 5]. The avidin-based approach provides several advantages: (1) the avidin helps to preventbinding of the recognition molecule to the surface and subsequent denatura-tion, (2) the number of linkages of the recognition molecule to the surfacecan be limited by controlling the number (and possibly the position) of thebiotins attached, and (3) the same surface can be used for immobilizing a widevariety of recognition molecules. This latter advantage is particularly useful

11 Fluorescence-Based Optical Biosensors 201

for making arrays of different recognition molecules to capture multiple tar-gets [6,7]. However, the biotin-avidin approach may have disadvantages if therecognition molecules used for capture are very small; in this case, a spacermay be required between the biotin and the capture molecule to make surethat it extends out of the binding pocket on the avidin and into the solution.Another situation where immobilization via an avidin bridge may not be theoptimum approach is where the density of the capture molecule is critical forefficient binding – as has been documented with sugars [8] and antimicrobialpeptides [9].

Assays with biological recognition molecules for fluorescence biosensorsare configured in four basic formats: direct binding, sandwich, competition,and displacement assays. The direct binding format is by far the simplest toimplement as it involves only the capture molecule and the target. However,it only works if either the target is inherently fluorescent or if it has somehowbeen prelabeled, as is the case in many DNA hybridization assays. Direct-binding assays are often used with prelabeled target standards for severalreasons: to select the best capture molecules for a particular application bycharacterizing affinity and specificity of the binding function; to optimize theassay conditions for sensitivity, speed, and reproducibility; and to providepositive controls to monitor assay performance. A variation of the direct-binding assay utilized molecular beacons (reviewed by Yao et al. in [10]). Inthese assays, a fluorescence energy transfer donor is immobilized on one areaof a capture molecule, while an acceptor is immobilized on an area that isadjacent only when no target is bound. When the target is bound, the twofluorophores separate, and a positive signal is generated.

The sandwich assay is used for detection of targets with at least two bind-ing sites, including large molecules such as proteins, oligonucleotides, bacte-ria, and viruses. In the sandwich assay, the target binds the capture molecule(usually immobilized) and a fluorescent tracer molecule, usually in that order.The formation of the resultant fluorescent complex is measured, while free flu-orescent tracer molecules are either removed or optically excluded from thesensing region. Sandwich assays are described in the majority of publicationson optical biosensors.

For small molecules with only a single-binding site, the formation of such asandwich is not possible and either a competitive or displacement assay mustbe utilized (Fig. 11.1). Two versions of the competitive assay are widely used.In one version, the capture molecules are immobilized, labeled target is addedto the sample solution, and the labeled and unlabeled target molecules com-pete for the binding sites on the immobilized capture molecule. In the secondversion, a target molecule is immobilized, labeled antibody is added to thesample solution, and any target free in solution prevents the binding of thelabeled antibody to the immobilized target. In both these versions, the fluo-rescent signal at the surface decreases with the increase in target in the samplesolution. Although the fluorescence in solution can also be measured, the flu-orescent component in solution is generally in sufficiently high concentration

202 F.S. Ligler

(A) Sandwich

Target Alexafluor or Cy5-Dye

(B) Competitive (C) Displacement (D) Direct Binding

Fig. 11.1. Assay configurations. Four different types of assays are possible usingfluorescence biosensors, with variations such as the binding of antibodies to thesurface and labeling of antigen in competitive assays in a variation of the formatshown in (b) or the use of molecular beacons to provide a signal upon direct bindingof antigen in a variation of the format shown in (d). The figure was prepared byKim Sapsford, George Mason University

that the very small decreases caused by low concentrations of target are dif-ficult to measure. One way to avoid this problem is to use a displacementassay. In this configuration, the active sites of an immobilized monoclonal an-tibody are saturated with a fluorescent analog of the target molecule. Pulsesof sample are introduced over the surface in a continually flowing stream.When the stream contains the target, the fluorescent analog is displaced fromthe immobilized capture molecule and measured downstream. Thus a signalis generated that increases in proportion to the amount of target present.

Of the most commonly described recognition molecules, oligonucleotidesprovide the greatest sensitivity. The principle reason is that the target genescan be amplified prior to the detection reaction using a polymerase and se-lective primers. This amplification step not only produces more molecules forthe detection reaction, but also increases the ratio of target to backgroundmolecules in complex samples. Fluorescence biosensors have been developedthat simply measure the process of amplification using dyes that bind to in-creasing concentrations of double stranded DNA, assuming that only targetoligonucleotides are replicated. For more selective analysis, the products of theamplification have been measured using capillary electrophoresis [11] or hy-bridization to selective probes in solution or on a surface (reviews in [1,2,10]).

The biorecognition molecules most frequently used by the optical biosensorcommunity have been antibodies. Antibodies have excellent selectivity andare relatively stable – unless they are exposed to too much heat or harshchemicals. However, to find the most appropriate antibody for each appli-cation can be a time-consuming process. Genetic engineering techniques arebeing used to create a bigger range of specific-binding molecules. Advancesin this area include the development of single-chain antibodies, in which the


heavy and light chain regions of the active site are cloned and produced asa single peptide, [8] and the formation of libraries of shark and camelid ac-tive site peptides, which are naturally composed of a single chain. The latter,in particular, are very stable and usually recover from chemical or thermaldenaturation to recover binding function [12].

The disadvantage of using DNA or antibodies for biorecognition is thatthe user must know exactly what the target is prior to assay development.To expand the repertoire of detectable targets, particularly for toxins andpathogens, molecules that recognize families of targets have been exploitedas capture molecules (reviewed in [8]). Sugars and gangliosides have beenused with the idea of mimicking the surface of the mammalian cells targetedfor infection by pathogen. Antimicrobial peptides, used by amphibians andother organisms for protection against pathogens, have been immobilized andproven to bind families of bacteria and toxins. Other candidates for broaderspectrum target recognition include siderophores, phages, and combinatorialpeptides.

11.3 Displacement Immunosensors

Displacement immunosensors are some of the simplest optical biosensors inboth form and function. However, why they work is not straight forward. Thebasic idea is as follows: First, a monoclonal antibody is immobilized on a solidsurface (beads, porous membranes, capillary walls). Second, the active siteson the antibody are filled with a fluorescent version of the target of inter-est. Third, a continuous stream containing injected samples is passed overthe immobilized antibodies. (This can continue for days.) When the targetis present, it displaces a small proportion of the bound, fluorescent analog.The fluorescent molecules pass downstream and are measured using a sim-ple fluorimeter. Simple antibody kinetics do not explain the behavior: if thefluorescent analog were bound with the dissociation constants typically mea-sured in solution, it would come off in the flow within 20 or 30 min. If it isbound very tightly, it would not come off instantaneously in the presence oftarget – but it does. The general understanding is that the fluorescent analogis continually in the process of association and dissociation with the antibody.However, it tends to remain in the boundary layer at the solid surface, onlydiffusing very slowly into the bulk flow. When a target is present, it preventsthe rebinding of the labeled analog that is dissociated at that point in time.The result is that, depending on the actual antibody–target pair selected, anoperator may be able to add sample after sample and get a positive responsefor 50 or 60 target-containing samples in series (Fig. 11.2). The displacementimmunosensor is particularly useful for the detection of low molecular weighttargets such as explosives and drugs of abuse. Biosensors for these applicationswere commercially available for a time. Sensitivities of part per million even topart per billion have been obtained in assays that only require a minute or two

204 F.S. Ligler

Fig. 11.2. Flow immunosensor using beads as the solid support. Target displaceslabeled target analog from immobilized antibodies to generate positive fluorescencesignal downstream. Reprinted from Ligler and Taitt (2002) with permission fromElsevier

to perform. For a review, see Kusterbeck in [1, 2]. Kusterbeck and colleagueshave recently developed a flow immunosensor to detect unexploded ordinancein the ocean that has been tested mounted on an unmanned, underwatervehicle (personal communication).

11.4 Fiber Optic Biosensors

Fiber optic biosensors utilize two distinct assay configurations for signal gen-eration and measurement: the optrode configuration and the evanescent waveconfiguration. Both configurations rely on the same principle of total internalreflection (TIR) for light propagation and guiding. However, optrodes use thelight shining out the end of the fiber to generate a signal either at the dis-tal face of the fiber or in the medium near the fiber’s end, while evanescentwave sensors rely on the electromagnetic component of the reflected light atthe surface of the fiber core to excite only the signal events localized at thatsurface. The penetration depth of the light into the surrounding medium ismuch more restricted than for optrodes, while the surface area interrogated ismuch larger in comparison to optrodes of equal diameter. The result is thatevanescent wave biosensors require immobilization of the biological recogni-tion molecules onto the longitudinal surface of the optical fiber core, primarily


measure binding events, and are relatively immune from interferents in thebulk solution. Optrode-type sensors are adaptable to a much wider varietyof measurements and produce a higher power excitation light in the sensingarea, but the assays must be formatted to accommodate excitation light in thebulk solution and signal collection from a very small surface, i.e. the fiber tip.A history of the earliest applications of fiber optics for biosensor applicationscan be found in [13].

11.4.1 Fiber Optics for Biosensor Applications

Total internal reflectance (TIR) is observed at the interface between two di-electric media with different indices of refraction as described by Snell’s law:

n1

n2sinθ1 = sinθ2, (11.1)

where n1 and n2 are the refractive indices of the fiber optic core and thesurrounding medium, respectively. The angle of light incident through thecore of the optical fiber is represented by θ1 and the angle of either the lightrefracting into the surrounding medium or the internal reflection back intothe core is represented by θ2. Total internal reflection requires that n1 > n2

and occurs when the angle of incidence is greater than the critical angle, θc,defined as:

θc = sin−1

(n2

n1

). (11.2)

This parameter must be considered when designing any biosensor on the ba-sis of optical fibers. However, although Snell’s law describes the macroscopicoptical properties of waveguides, it does not account for the electromagneticcomponent of the reflected light, known as the evanescent wave. The evanes-cent wave is an electric field that extends from the fiber surface into the lowerindex medium and decays exponentially with distance from the surface, gener-ally over a distance of hundred to several hundred nanometers. For multimodewaveguides, the penetration depth, dp, the distance at which the strength ofthe evanescent wave is 1/e of its value at the surface, is approximated by:

dp =λ

(4π[n21sin

2θ − n22]1/2)

(11.3)

where n1 and n2 are refractive indices of the optical fiber and surroundingmedium, respectively, and θ is the angle of incidence [14].

The importance of the evanescent wave is its ability to couple light outof the fiber into the surrounding medium, thereby providing excitation forfluorophores bound to or in proximity to the fiber core surface. This confinedrange of excitation is one of the major factors responsible for the relative im-munity of evanescent wave-based systems to the effects of matrix componentsor interferents beyond the reaction surface (Fig. 11.3). Love and Button [15]

206 F.S. Ligler

Fig. 11.3. Evanescent illumination of fluorescent complexes at the surface of thecore of a partially clad waveguide. Reprinted from Ligler and Taitt (2002) withpermission from Elsevier

Fig. 11.4. Strategies for separation of excitation and emission light paths (Liglerand Taitt, 2002). Figure reprinted with permission from Elsevier

originally suggested that dipoles close to the surface could emit approximately2% of their radiated power (fluorescence) into modes coupled back up the fiber;however, Polercky et al. (2000) have more recently analyzed films of dipolesand concluded that a much higher proportion of the radiation can couple intoguided modes.

A variety of configurations have been utilized with optical fibers for excit-ing fluorescence and collecting the emitted signal. In Fig. 11.4a, a traditionaldichroic mirror is used to separate excitation and emission on the basis of thedifference in wavelengths. The excitation light may pass through the mirror,while the emission light is reflected onto a detector (left) or, in a scheme thathas generally proven more effective, the stronger excitation light is reflectedonto the end of the fiber, while the emitted light passes straight through. InFig. 11.4b, the excitation light passes through a hole in an off-axis parabolic


mirror, while the emission light is reflected by the mirror onto a detector [16].In Fig. 11.4c, a fiber bundle is used between a high numerical aperture sens-ing probe and the optics. The 635 nm excitation light is channeled down thecentral silica fiber while the emitted higher wavelength fluorescence is coupledback up the surrounding plastic fibers [17]. In Fig. 11.4d, the fiber is illumi-nated using a light source normal to the fiber, and the fluorescence is detectedat the distal end [18,19].

11.4.2 Optrode Biosensors

Optrode biosensors grew out of a wealth of applications for chemical sensingon the basis of absorbance, fluorescence, and even time-resolved fluorescence.The fibers were relatively easy to work with and facilitated the separation ofexcitation and emission light; frequently, the fibers themselves were used totransmit only excitation to the sensing region or only emission away from thesensing area as shown in the diagram below. Furthermore, a wide variety oflight sources and detectors were used. These included lamps, LEDs, lasers,and diode lasers for excitation and photomultiplier tubes, photodiodes, andCCDs for detection. The optrode configuration is one of the most flexiblein terms of the types of assays that can be performed. Optrodes have beenused not only with immunoassays and DNA hybridization assays, but alsowith enzyme assays and for analysis of whole cell function. These assays andconfigurations have been reviewed by Walt [1, 2] (Fig. 11.5).

Probably the most exciting application of optrode technology today is thefiber optic bundles developed by Walt [20] and currently marked by Illumina(www.Illumina.com). In this approach, the ends of the optical fibers in thebundle are etched out, leaving the surrounding cladding to form microwells.Coded beads are placed in the microwells so that hundreds of thousands of as-says can be conducted simultaneously. The assays are being used for genomics,proteomics, and biodefense.

11.4.3 Evanescent Fiber Optic Biosensors

The first use of an optical fiber as a biosensor, and one of the first opticalbiosensors was described by Hirschfeld and Block [21, 22] in the mid 1980s.This type of sensor was also the first biosensor to be fully automated and usedremotely. In 1996–1998, Ligler, Anderson, and colleagues developed a biosen-sor payload for a small, unmanned plane and tested it at Dugway ProvingGround, Utah. The payload included a biosensor with four fiber probes, aRAM-air driven cyclone for collecting aerosolized bacteria, an automated flu-idics system, batteries, and a radiotransmitter. They demonstrated that theairborne biosensor could collect a biothreat stimulant, identify it, and radiothe data to an operator on the ground in 6–10 min [23,24].

The best known of the commercially available fiber optic biosensors is theRAPTOR-Plus, made by Research International (http://www.resrchintl.com)

208 F.S. Ligler

Light source

Detector

Detector

Detector

Sensing layer

Light source

Light source

(a)

(b)

(c)

Fig. 11.5. Design principles for optrode biosensors. (a) Two fibers: one carries lightto the sensing layer and one carries the signal to the detector. (b) Bifurcated fiber:the biosensing layer is placed on the fused end of the fiber. (c) The biosensing layeris placed on the central fiber, and the surrounding fibers are used to collect the lightsignals. Reprinted from [1] with permission from Elsevier

Fig. 11.6. Schematic (a) and photograph (b) of injection-molded optical fiberprobes and fluidics coupon for use with the RAPTOR-Plus. (a and b from [1],reprinted with permission from Elsevier). (c) Photograph of the Raptor Plus withkeypad control and result window on the left and the sample port and reagentchamber on the right

(Fig. 11.6). This system is proving to be very reliable in terms of long termoperation (¿3 years to date, George Anderson, personal communication). Inboth the 4-probe Raptor and the more recently developed 8-fiber BioHawk,a disposable coupon contains the fiber probes, providing protection dur-ing long-term storage and fluidic channels for automated sample processing.


The polystyrene probes are inserted after being coated with the recognitionbiomolecule and dried. The probes are designed to integrate a combinationtapered sensing region with a lens for signal [25]. The coupon automaticallyaligns the probe so that it is in the middle of the fluid channel [26]. In theRaptor, the light emitted from the end of the fiber is focused onto a collec-tion lens in the permanent portion of the biosensor, while in the BioHawk,the light is collected normal to the fiber. The Raptor has been used to assayfor a wide variety of targets, including explosives, toxins, clinical markers,and pathogens. It has proven to have sensitivities comparable to traditionalenzyme-linked immunoassays (ELISA) using the same recognition molecules,but with assay times in the 10–15 min range instead of 2–4 h. Furthermore,the assays can be performed in the presence of highly complex sample matri-ces, including blood, urine, food homogenates, beverages, ground water, andeffluents from air samplers. Results of such assays have been reviewed in [1,2].

11.5 Bead-Based Biosensors

The use of beads as biosensors was first made popular by Raoul Kopelmanwith his PEBBLES (reviewed by Brasuel et al. in [1]). PEBBLES were smallbeads originally including ion- or pH-sensitive dyes that could be injectedinto cells to measure local microenvironments. Imaging was performed usinga fluorescent microscope. The Kopelman group expanded the repertoire ofPEBBLES to include the use of more specific recognition elements such ascalcium ionophores for real-time interrogation of cell function.

Building on this approach, other groups developed highly specializednanoparticles to measure other intracellular functions. One of the more inter-esting particles is composed of quantum dots modified with binding proteinand a quenching dye [27]. In this system, a dye molecule is positioned on thebinding protein adjacent to the active site. The dye quenches the signal fromthe quantum dot unless the binding protein binds its target. At that point, theenvironment of the dye changes so that it can no longer quench the quantumdot, and a positive signal is generated. Medintz et al. used a broad-spectrumquenching dye coupled to genetically modified maltose-binding proteins anddemonstrated the capacity of different colored dots to generate signals in re-sponse to different target analytes.

The next major breakthrough in bead-based sensing was the developmentof coded beads that enabled simultaneous measurement of multiple analytesusing beads with different amounts of two dyes. Originally developed for com-binatorial chemistry applications, the fluorescent coded beads enabled highlymultiplexed assays to be performed in a single fluid compartment. The bun-dled optrode was perfect for making such measurements (see Yu and Waltin [2]) as was flow cytometry [28]. In both the cases, beads of each code ordye ratio are coated with a distinct biorecognition molecule. When the tar-get binds to the bead of that code, another label is introduced using either

210 F.S. Ligler

Fig. 11.7. High-density multianalyte bio-optrode composed of microsphere array onan imaging fiber. (a) Scanning force micrograph (SFM) of microwell array fabricatedby selectively etching the cores of the individual fibers composing the imaging fiber.(b) The sensing microspheres are distributed in the microwell. (c) Fluorescence im-age of a DNA sensor array with ∼13,000 DNA probe microspheres. (d) Small regionof the array showing the different fluorescence responses obtained from the differ-ent sensing microspheres [20]. Reprinted from [1] with permission from AmericanAssociation for the Advancement of Science

a sandwich reaction or a third die, e.g., one that binds to double strandedDNA. Then only the beads with target have three colors and the identity canbe read from the bead code (Fig. 11.7).

11.6 Planar Biosensors

Planar waveguide biosensors also usually employ evanescent illumination froma coherent source to excite fluorophores at the surface of the waveguide. Theresulting fluorescence emission can be measured either by a single photode-tector or by a system capable of imaging, usually a CCD or CMOS camera.A variety of focusing lenses have been used to improve detector response[7, 17,29–32]. The introduction of bandpass and longpass filters was found toimprove the rejection of scattered laser light and hence reduce the backgroundof the system [7] (Fig. 11.8).

Unfortunately, a side effect of using bulk waveguides and collimated lightis the production of sensing “hot spots” along the planar surface that occurwhere the light beam is reflected, illuminating only discrete regions. These


Flow Chamber

Waveguide Diode Laser

GlassWindow

Emission Filter

GRIN Lens Array

CCD Imaging Array

Fig. 11.8. Detection of emitted fluorescence at right angles to the waveguide inter-face as described by Golden et al., 1999. Reprinted with permission of the Elsevier

hot spots have been successfully utilized as sensing regions by Brecht andcoworkers in the development of an immunofluorescence sensor for wateranalysis [33,34]. Feldstein et al. [7,45] overcame this problem by incorporatinga line generator and a cylindrical lens to focus the beam into a multimodewaveguide, which included a propagation and distribution region prior to thesensing surface. This resulted in uniform lateral and longitudinal excitationat the sensing region.

Another method of achieving uniform longitudinal excitation of the sensingregion is to decrease the waveguide thickness [30]. When the thickness of thewaveguide is much greater than the wavelength of the reflected light, thewaveguide is referred to as an internal reflection element (IRE). However,if the thickness of the waveguide is decreased such that it approaches thewavelength of the incident light, and the path-length between the points oftotal internal reflection become increasingly shorter. At the thickness wherethe standing waves, created at each point of reflection, overlap and interferewith one another, a continuous streak of light appears across the waveguide,and the IRE becomes known as an integrated optical waveguide (IOW) [35].

Integrated optical waveguides, used frequently in TIRF studies, are mono-mode and prepared by depositing a thin film of high refractive index materialonto the surface of a glass substrate. These thin films are typically 80–160 nmin thickness and consist of inorganic metal oxide compounds such as tin oxide[29], indium tin oxide [36], silicon oxynitride [37], and tantalium pentoxide [38,39]. The light is coupled into these IOWs via a prism or grating arrangement.Studies by Brecht and coworkers compared IRE and IOW-based waveguidesand concluded that the integrated optics significantly improved the sensitivityof the system by a factor of 100 [33].

212 F.S. Ligler

anti-

SEB

anti-

B.a

nthr

acis

anti-

B.a

bortu

san

ti-F.

tula

rens

isan

ti-C

txan

ti-R

icin

100 ng/ml SEB

7.1x104 cfu/ml B. anthracis Sterne

1.4x105 cfu/ml killed B. abortus

7.3x106 cfu/ml F. tularensis LVS

200 ng/ml Ricin

100 ng/ml Cholera toxin

Direction of flow

Fig. 11.9. Antibody array used to interrogate six samples for six different biothreatagents. Reprinted from [40] with permission of the Elsevier

The planar geometry provides a surface for depositing arrays of detectionmolecules. This makes it ideal for multianalyte detection. An example of suchan array is shown below. In numerous papers reviewed by Sapsford et al.in [1,2], the limits of detection are typically about an order of magnitude bet-ter with good array biosensors, based on planar waveguides than for standardELISAs using the same reagents. In addition to antibodies and DNA, detec-tion molecules used with planar waveguide biosensors include siderophores,carbohydrates, antimicrobial peptides, and gangliosides [2] (Fig. 11.9).

11.7 Critical Issues and Future Opportunities

Fluorescence-based optical biosensors have inherent advantages for discrimi-nating targets in complex samples and for increasing sensitivity compared withlabel-free methods. Limits of detection can be further reduced using amplifi-cation methods such as enzymes, polymerases, or multifluorophore complexes(e.g., labeled dendrimers as in [41], or virus particles carrying 60 carefully sep-arated fluorophores as in Sapsford et al., 2006). However, labeling and ampli-fication procedures can complicate assay processes, making automation moredifficult and possibly increasing fluorescence background. Another methodfor increasing signal include preconcentration of target prior to analysis using


filters, dielectrophoreses, or magnetic bead separations. The latter approachhas been very successfully implemented in commercial systems for chemilu-minescence assays (reviewed by Richter [1, 2]).

Microfluidics are becoming increasingly used for three particular facets ofoptical biosensors:

1. Microfluidic devices facilitate methods for automated sample processing,including the mixture of the sample with the fluorescent reagents.

2. They provide an efficient mechanism for manipulating very small volumes,which saves reagent costs.

3. They can deliver the target more efficiently to the sensing surface.

Examples of the latter are laminar fluid focusing as described by Hofmannet al. [42] and Munson et al. [43] and passive mixing to minimize target dep-letion at the sensing surface as described by Golden et al. [44]. However,care must be taken to make sure that fluidic channels are sufficiently largeto avoid clogging by complex samples, and that they have the appropriatesurface chemistry to prevent nonspecific adsorption of the target outside thesensing region. Furthermore, the issue of interrogating a sufficient volume toinclude targets present at low concentrations must never be neglected.

Advances in optics offer new opportunities for increased sensitivity andreductions in size and cost. Silicon technology is producing better and betterintegrated optical waveguides for highly multiplexed analysis. Arrays of sin-gle photon detectors are described by Eduardo Charbon in this volume thatmay offer the opportunity to detect single targets if the affinity of the recog-nition molecule is sufficiently high and if the background can be sufficientlyreduced. Both scattered excitation light and stray fluorescence from moleculesnot bound in the detection complex can generate “noise” beyond that inher-ent in the optical device itself. The production of devices based on organicpolymers is also very exciting because they should be relatively simple to inte-grate biological recognition elements and polymer-based microfluidics to formmonolithic, inexpensive, or disposable sensors. Organic LEDs, transistors, andphotodiodes are described in this volume in the chapter by Peter Seitz.

Finally, systems biology at the molecular level is identifying new targetsfor analysis that are of importance for medical, environmental, and defenseapplications. Furthermore, the understanding of how to make rationally de-signed molecules for sensing applications offers new approaches to performthe biorecognition function with appropriate specificity, increased sensitivity,and enhanced stability during sensor storage and use. New fluorophores alsoprovide opportunities for interrogating new types of samples and systems andfor generating multiple signals simultaneously.

The future for fluorescence-based optical biosensors will be rich. If thereis a real limitation, it is only on the ability to synthesize all the emergingtechnologies into the most useful system for each customer. That ability restson the willingness of scientists and engineers in universities, government, andindustry to work together in interdisciplinary teams and the longer term visionof those that support such efforts.

214 F.S. Ligler

Acknowledgments

The preparation of this chapter was supported by NRL 6.2 Work Unit 6006.The views expressed herein are those of the author and do not representopinion or policy of the US Navy or Department of Defense.

References

1. Ligler, F.S. and C.R. Taitt: Optical Biosensors: Today and Tomorrow, (Elsevier,2002).

2. Ligler, F.S. and C.R. Taitt: Optical Biosensors 2008: Present and Future,(Elsevier, 2008).

3. Cass, A.T. and F.S. Ligler: Immobilized Biomolecules in Analysis: A PracticalApproach, (Oxford University Press, Oxford, UK 1998).

4. King, K.K., J.M. Vanniere, J.L. LeBland, K.E. Bullock, and G.P. Anderson:Environ. Sci. Technol. 34, 2845 (2000).

5. Herron, J.N, S. zumBrunnen, J.-X. Wang, X.-L. Gao, H.-K.Wang, A.H. Terryand D.A. Christensen: Proc. SPIE 3913, 177 (2000).

6. Silzel, J. W., B. Cercek, C. Dodson, T. Tsay and R. J. Obremski: Clin. Chem.44, 2036 (1998).

7. Feldstein, M. J., J. P. Golden, C. A. Rowe, B. D. MacCraith and F. S. Ligler:J. Biomed. Microdevices 1:2, 139 (1999).

8. Ngundi, M.M., C.R. Taitt, S.A. McMurry, D. Kahne, and F.S. Ligler Biosens.Bioelectron. 21, 1195 (2006).

9. Kulagina, N.V. K.M. Shaffer, G.P. Anderson, F.S. Ligler, and C.R. Taitt: Anal.Chim. Acta 575, 9 (2006).

10. Thompson, R. B. In: Fluorescence Sensors and Biosensors, ed by Taylor &Francis, (Boca Raton, FL, 2005), p. 394

11. Lagally, E.T., J.R. Scherer, R.G. Blazej et al : Anal. Chem. 76, 3162 (2004).12. Goldman, E.R., G.P. Anderson, J.L. Liu, J.B. Delehanty, L.J. Sherwood, L.E.

Osborn, L.B. Cummins, and A. Hayhurst: Anal. Chem. 79, in press (2007).13. Wise, D.L., and L.B. Wingard, Jr.: Biosensors with Fiberoptics, (Human Press,

NJ, 1991), p. 37014. Harrick, N.J.: Internal Reflection Spectrosopy, (Wiley, New York 1967), p. 32715. Love, W.F. and L.J. Button: Proc. SPIE 990, 175 (1988).16. Thompson, R.B. and M. Levine: US Patent No. 5,141,312 (1992).17. Wadkins, R.M., J.P. Golden and F.S. Ligler: J. Biomed. Optics 2, 74 (1997).18. Kooyman, R.P.H., H.E. de Bruijn and J. Greve: 1987, Proc. SPIE 798, 290

(1987).19. Ligler, F.S., M. Breimer, J.P. Golden, D.A. Nivens, J.P. Dodson, T.M. Greene,

D.P. Haders and O.A. Sadik: Anal. Chem. 74, 713 (2002).20. Walt, D. R.: Science 287, 451 (2000).21. Hirschfeld, T.: US Patent No. 4,447,546 (1984).22. Hirschfeld, T.E. and M.J. Block: US Patent No. 4,558,014 (1985).23. Ligler, F.S., G. P. Anderson, P.T. Davidson, R.J. Foch, J.T. Ives, K.D. King,

G. Page, D.A. Stenger and J.P. Whelan: Environ. Sci. Technol. 32, 2461 (1998).


24. Anderson, G.P., K.D. King, D.S. Cuttino, J.P. Whelan, F.S. Ligler,J.F. MacKrell, C.S. Bovais, D.K. Indyke and R.J. Foch: Field Anal. Chem.Technol. 3, 307 (1999).

25. Saaski, E.W. and C.C. Jung: US Patent No. 6,136,611 (2000).26. Saaski, E.W.: US Patent No. 6,082,185 (2000).27. Clapp, A.R., I.L. Medintz, H.Tetsuo Uyeda, B.R.Fisher, E.R.Goldman, M.G.

Bawendi, and H.Mattoussi: J. Amer. Chem. Soc. 127, 18212 (2005).28. Edwards, B.S., T. Oprea, E.R. Prossnitz, L.A.Sklar: Current Opinion in Chem-

ical Biology 8, 392 (2004).29. Duveneck, G. L., D. Neuschafer and M. Ehrat: International Patent Go1N 21/77,

21/64 (1995).30. Herron, J. N., D. A. Christensen, K. D. Caldwell, V. Janatova, S.-C. Huang and

H.-K. Wang: US Patent 5,512,492 (1996).31. Herron, J. N., D. A. Christensen, H.-K. Wang and K. D. Caldwell: US Patent

5,677,196 (1997).32. Golden, J. P.: US Patent 5,827,748 (1998).33. Brecht, A., A. Klotz, C. Barzen, G. Gauglitz, R. D. Harris, G. R. Quigley, J. S.

Wilkinson, P. Sztajnbok, R. Abuknesha, J. Gascon, A. Oubina and D. Barcelo:Anal. Chim. Acta 362, 69 (1998).

34. Koltz, A., A. Brecht, C. Barzen, G. Gauglitz, R. D. Harris, G. R. Quigley, J. S.Wilkinson and R. A. Abuknesha, 1998, Sens. Actuators B 51, 181.

35. Plowman, T. E., S. S. Saavedra and W. H. Reichert: Biomaterials 19, 341 (1998).36. Asanov, A. N., W. W. Wilson and P. B. Oldham: Anal. Chem. 70, 1156 (1998).37. Plowman, T. E., J. D. Durstchi, H. K. Wang, D. A. Christensen, J. N. Herron

and W. M. Reichert: Anal. Chem. 71, 4344 (1999).38. Duveneck, G. L., M. Pawlak, D. Neuschafer, E. Bar, W. Budach, U. Pieles and

M. Ehrat: Sens. Actuators B 38-39, 88 (1997).39. Pawlak, M., E. Grell, E. Schick, D. Anselmetti and M. Ehrat: Faraday Discuss.

111, 273 (1998).40. Rowe-Taitt, C. A., J. P. Golden, M. J. Feldstein, J. J. Cas, K. E. Hoffman and

F. S. Ligler: Biosens. Bioelectron. 14, 785 (2000).41. Chang, A.-C., J.B. Gillespie, and M.B. Tabacco: Anal. Chem. 73, 467 (2001).42. Hofmann, O., G.Voirin, P. Niedermann, and A. Manz: Anal. Chem. 74, 5243

(2002).43. Munson, M.S., M.S. Hasenbank, E. Fu, and P. Yager: Lab on a Chip 4, 438

(2004).44. J.P. Golden, T. Floyd-Smith, and F.S. Ligler: Biosens. Bioelectron., in press

(2007).45. Feldstein, M. J., B. D. MacCraith and F. S. Ligler: US Patent 6,137,117 (2000).46. Golden, J.P., E.W. Saaski, L.C. Shriver-Lake, G.P. Anderson and F.S. Ligler:

Opt. Eng. 36, 1008 (1997).47. Hirschfeld, T.E.: and M.J. Block: US Patent No. 4,447,546 (1984)48. Ngundi, M.M., N.V. Kulagina, G.P. Anderson, and C.R. Taitt: Expert Rev.

Proteonomics 3, 511 (2006).49. Polerecky, L., J. Hamrie, and B.D. MacCraith: Appl. Opt. 39, 3968 (2000).50. Sapsford, K.E., P.T. Charles, C.H. Patterson Jr., and F.S. Ligler: Anal. Chem.

74, 1061 (2002).51. Sapsford, K.E. , C.M. Soto, A.S. Blum, A. Chatterji, T. Lin, J.E. Johnson, F.S.

Ligler and B.R. Ratna: Biosens. Bioelectron. 21, 1668 (2006).

12

Optical Biochips

P. Seitz

The citizens of our modern society are exposed to an increasing number ofchemical substances that are potentially harmful or that need to be monitoredand kept under check. At the same time, everybody would like to know muchearlier, with higher certainty and at lower cost, whether her or his health isaffected in any way. For this reason there is a huge demand for very sensitive,highly specific, cost-effective, and rapid methods to detect the concentrationof bio-active molecules. Because of the simplicity, efficiency, and speed withwhich light can be generated, manipulated, detected, and used to sense the ef-fects of many chemical reactions, optical techniques have become the methodof choice for many sensing problems involving bio-active molecules. A par-ticularly attractive aspect is the possibility to miniaturize and to parallelizeoptical measurements, which has led to the very active field of integratedoptical sensing [1]. Such miniaturized optical systems for the sensing of bio-active molecules are also known as biochips, although this name is used formany different levels of integration. Applications of these optical biochips in-clude medical diagnostics (for specialized laboratories and increasingly alsofor home use), contaminant detection in the food industry (e.g., hormonesin milk or meat), pharmaceutical research and development (drug screening),environmental and pollution monitoring (e.g., pesticides in water), securityand counter-terrorism (detection of chemical and biological warfare agents),as well as process and quality control in industry.

The present contribution reviews the different principles and realizations ofoptical biochips, and it offers an outlook on the potential of monolithically in-tegrating complete optical measurement systems on one single, self-containedbiochip of unprecedented complexity and functionality.

12.1 Taxonomy of Optical Biochips

12.1.1 Basic Architecture of Optical Biochips

The principle architecture of an optical measurement system for bio-activemolecules is illustrated in Fig. 12.1. Coordinated by an electronic control

218 P. Seitz

Fig. 12.1. Principle architecture of an optical measurement system for chemicaland biochemical sensing, as used for the different types of optical biochips

system, which can be a single microcontroller chip in the simplest case, dif-ferent optoelectronic, optomechanic, and microfluidic components interact todetect the effects of a chemical reaction, which is sensed by an optical effect inthe system’s transducer section [2]. A light source is emitting either modulatedor unmodulated light, which is conditioned by input optics such as focusing,in-coupling, or angle-adjustment elements. In the transducer part, a chemicalreaction takes place between the analyte (the target molecule) and a suitablereceptor in the microfluidics part of the system. This reaction is observed bymeasuring a suitable change of any of the optical properties in the fluid vol-ume where the reaction is taking place. The light exiting the interaction regionis carrying the information about the analyte concentration, and it needs tobe conditioned by appropriate output optics, such as imaging, out-coupling,or angle-adjustment elements, so that the light can be efficiently detected bya photosensor. The resulting electric signal is acquired and converted by anelectronic analog/digital circuit, and the digital signal is the processed to ex-tract and communicate the desired concentration information, for example,by displaying it for the user on a screen.

Although it is possible to employ many optical effects (see Sect. 12.3 below)in an extended volume of the transducer part, much more control can be exer-cised if this interaction volume is restricted to the immediate vicinity (typicallyless than 1 µm) of a sensitized surface. This principle is illustrated in Fig. 12.2.The target molecules together with other possibly bio-active molecules are insolution, and the microfluidics system has the task to transport them to andfrom the transducer volume. There, receptor molecules are chemically bondedto a surface where light is passing to sense the effect of the chemical bonding.These receptors should react in a highly specific way only with the analyte, sothat the analyte molecules remain attached to the surface, effectively changingthe mass density close to the surface, which can be subsequently measured.Since no additional measure is necessary to determine the concentration ofthe analyte, this approach is called label-free sensing.

A higher detection sensitivity can be obtained if the analytes are taggedwith photoactive molecules. Most often, these labels are fluorescent molecules

12 Optical Biochips 219

Fig. 12.2. Illustration of the transducer principles in an optical biochip. Left : Label-free sensing principle. Right : Labeled sensing principle

that emit light when excited with radiation of shorter wavelength. Alterna-tively, one can also make use of luminescent molecules, such as the famousluciferase, which are causing a biochemical enzymatic reaction as part of which(visible) light is emitted [3]. The term “biochip” is used very broadly in thefield and does not describe the particular implementation of the principlesdescribed above precisely. Depending on the level of integration, one distin-guishes usually between three different types of biochips:

Biochemical Microarray

The simplest and still the dominating type of biochip consists of a one- or two-dimensional array of transducer regions, which are sensitized with differentreceptors on a common substrate [4]. Neither microfluidics transportationsystems nor the light sources or the detectors are part of such a biochemicalmicroarray. This type of a biochip can consist, therefore, of up to one millionindividual transducer regions, each with an area as small as 5× 5 µm2, on anoptical biochip with an active surface of 1–2 cm2 [5]. Most often, biochemicalmicroarrays employ fluorescence tagging as the optical transducer mechanism.If they are employed for DNA analysis, optical microarrays are known as DNA-chips or GenechipsTM, and such DNA-chips with half a million transducerregions are commercially available for about ¤200 [5]. Although DNA analysis(genomics) is still the predominant application of optical microarrays, they areincreasingly used for protein (proteomics) and cell analysis [6].

Lab-on-a-Chip (µ-TAS)

To simplify the practical use of biochips and to reduce the sample volumesrequired for an analysis, an increasing amount of functionality is placed on thesame substrate. A lab-on-a-chip (LOC) combines the microfluidic sub-system,including all elements required to transport, mix, process, separate, and drainfluids, together with the optical transducer part on a single chip [7]. Such an

220 P. Seitz

Fig. 12.3. Building blocks of an integrated diagnostic-therapeutic biochip

LOC can analyze minute sample volumes measured in femtoliters (equivalentto µm3). In 1990, researchers at Ciba-Geigy (Basel, Switzerland) coined thealternate name Micro Total Analysis System (µ-TAS) for an LOC [8].

Since biochips are disposable items, many LOCs operate with externallight sources, imaging optics, and 1D or 2D photosensor arrays (solid-statecameras). However, first LOCs have been realized making use of integratedlight sources and/or photosensors, as, for example, in the OSAILS (opticalsensor array and integrated light source) principle [9].

Diagnostic-Therapeutic Microsystems

The ultimate biochip consists not only of a complete LOC to perform a com-prehensive analysis of biological, chemical, and physical parameters, it alsoincludes a therapeutic BioMEMS (Micro-Electro-Mechanical System), suchas a drug delivery or an electro-stimulation sub-system [10] (Fig. 12.3). Beforesuch a vision of an integral biochip becomes reality, many practical problemshave to be solved, such as the regeneration of biochemical sensing surfaces,bio-fouling resistance, robustness in the chemically aggressive environment ofthe body of animals and human beings, long-term power-supply using bio-fuel harvesting or inductive coupling, as well as reliable communications withother implanted systems or the outside world [11].

12.2 Analyte Classes for Optical Biochips

The analytes in our optical biochips were simply illustrated as generic mole-cules in Fig. 12.2. Depending on the actual biochemical function of the typeof molecule, different classes of analytes are distinguished [4].

12.2.1 DNA (DNA Fragments, mRNA, cDNA)

To map or sequence the genetic information of a cell, DNA chips contain alarge number (up to one million) of different single-stranded DNA fragments


(so-called oligonucleotides), which are immobilized on the substrate surface indiscrete spots. Each spot contains several millions of identical oligonucleotidesto increase sensitivity. The sample to be tested contains single-stranded ge-netic chains (DNA fragments, mRNA, or cDNA), and they are usually labeledwith a fluorescent dye. The genetic chains that match the immobilized oligonu-cleotides bind to the spots on the substrate. The biochip is then illuminatedwith a suitable wavelength, so that the fluorescence light pattern of the dif-ferent spots allows the determination of the type and concentration of targetgenetic chains in the sample.

12.2.2 Proteins (Antigens)

Since the genetic information contains the instructions that proteins are pro-duced in the body, the identification of proteins – so-called proteomics – iseven more important than genomics. For this reason, the utilization of biochipsfor protein analysis is growing at a rate of about 29%, while the growth ofDNA-chips is only about 18% [12].

A protein biochip is similar to a DNA-chip, but instead of oligonucleotides,protein probes (antibodies) for the target proteins (antigens) are immobilizedon the individual spots on the biochip’s surface. However, protein biochipsare more difficult to handle, less specific, and less sensitive than DNA-chipsfor several reasons [4]: Proteins are unstable, and they can easily be dena-tured at solid–liquid and liquid–air interfaces. Today’s production techniquesfor protein probes produce only low-affinity capture antibodies, making thecoupling of the antigens to the antibodies not very specific. Since antibodieshave usually rather large surface areas for interaction, they show importantcross-reactivity between target proteins. DNA samples can readily be am-plified using PCR (polymerase chain reaction), while no such amplificationprocess is known for proteins.

The optical principles employed in protein biochips include fluorescence aswell as evanescent wave sensing (see Sect. 12.4).

12.2.3 Specific Organic Molecules

The same principle used for DNA and protein biochips can be employed, inprinciple, to determine the concentration of various organic molecules that areof biochemical importance. Suitable probe molecules are fixed on the surface ofa biochip. Corresponding target molecules then attach to the probe molecules,and their presence can be measured using the same optical techniques as usedfor the other types of biochips.

12.2.4 Cell Gene Products (cDNA, Proteins)

A cell microarray consists of a number of spots with different types of geneticinformation, such as DNA strands, plasmids, or adeno viruses. Mammalian

222 P. Seitz

cells are grown on these spots, where they can absorb the prepared geneticinformation – the cells are being transfected – and they are expressing the cor-responding cDNA molecules. This genetic information, in turn, is employedby the cell to produce a specific protein. This protein can influence the func-tioning of the live cell, and various experiments can then be carried out tostudy the role of the particular protein in the cell’s metabolism.

After the particular proteins are expressed in a cell, the cell can be in-cubated with fluorescently labeled antibodies, and the protein presence andconcentration can be measured and imaged with fluorescence microscopy.

Although cell biochips are only useful for cells that are easily transfected,they overcome many of the deficiencies of conventional protein biochips de-scribed earlier, and cell biochips are adding pertinent information about thecell’s metabolism to our knowledge [13].

12.2.5 Tissue

Instead of preparing a two-dimensional pattern of receptor molecules on abiochip, one can assemble a tissue microarray by combining tissue samplesfrom a large number of specimens on the same substrate that can then beanalyzed in a highly parallel fashion, for example, for specific protein expres-sion patterns. The typical diameter of an individual tissue spot on a tissuemicroarray is 600 µm and its thickness is about 5–8 µm.

A particularly important application of such tissue microarrays is thehigh-throughput molecular profiling of tumors [14]. This technology is alreadyoffered commercially, and other applications are seen for large-scale epidemi-ology studies, for the development of new diagnostic and prognostics markersfor tumors, for the investigation of biochemical pathways, for drug develop-ment, and for quality control of food, in particular, if genetically modifiedfoods are involved [4].

12.3 Optical Effects for Biochemical Sensors

The many ways in which light interacts with matter can all be exploited forthe detection of the concentration of an analyte by making use of suitablechemical reaction between the analyte and a receptor (or detector) molecule,either in the bulk or at the surface of an optical biochemical sensor. Thetaxonomy of these different optical effects that may be used for biochemicalsensing is illustrated in Fig. 12.4, see also [1–4]. In the following, the mostpopular of these optical effects are briefly described.

12.3.1 Spectral Absorption

For a long time, colorimetric sensing principles have been used very widelybecause the visual color perception of the observer could be exploited, and


Fig. 12.4. Taxonomy of optical effects potentially useful for biochemical sensing

no additional instrument was necessary. These optical sensing methods werebased on chemical reactions of the analyte with a receptor molecule thatresulted in a chemical product with optical absorption properties in the visiblespectral range. The resulting color change was then detected and evaluated bythe observer or an instrument. Since optical measurement techniques based onspectral absorption are less sensitive than methods detecting phase changesin light, optical biochips do generally not make use of spectral absorption.

12.3.2 Phase Shift

A change in the real part of the refractive index causes a phase change ofthe electromagnetic wave that is used to probe the detection volume of thesensor. This phase change can either be observed in an interferometer setup,it can manifest itself as a change of the polarization properties of incidentlinearly polarized light, it can change the optical field distribution close to aninterface, or it can change the propagation characteristics of a traveling wave.This last effect is most commonly exploited in optical biochips, in particular,through the modification of in- or out-coupling conditions, because it worksvery well in the confining geometries at the surface of biochips, as illustratedin Fig. 12.2, and it does not require active labeling of the analyte molecule.

12.3.3 Fluorescence

Because of the extremely high sensitivity of today’s solid-state image sensors,the detection of very low light levels down to individual photons is technicallynot difficult to achieve [15]. As a consequence, the use of fluorescent labels totag analyte molecules leads to a highly sensitive optical measurement methodfor the concentration of the analyte. Incident light at a certain wavelength isabsorbed by the fluorescent labels, and the optical energy is re-emitted at alonger wavelength where it is detected by a sensitive camera. For this reason,fluorescence is one of the most widely used sensing methods in the life sciencesand, in particular, in optical biochips.

12.3.4 Luminescence

Instead of attaching a fluorescent molecule to the analyte, it is also possible tobond a chemiluminescent molecule to it, such as the well-known luciferase [3].

224 P. Seitz

The resulting enzymatic chemical reaction leads to the emission of (visible)light, without any need for an optical excitation, and this light can be detectedby a sensitive camera. Since it is often easier to label an analyte with a fluo-rescent molecule than to label it with a suitable chemiluminescent molecule,luminescence is not very commonly used in optical biochips.

12.3.5 Raman Scattering

Raman scattering describes an optical process in which incident light of acertain energy is absorbed by a molecule, and it is re-emitted with a slightlydifferent energy, where the energy difference is a characteristic function of thechemical structure of the molecule. Although Raman scattering is very specificand rich in the information it carries, its efficiency is very low: Typically, onlyabout one in 105 photons is Raman scattered [4]. For this reason, Ramanscattering is currently not used in optical biochips.

12.3.6 Nonlinear Optical (NLO) Effects

If the electrical field strength is not too high, the optical polarization of amolecule is linearly proportional to the amplitude of the applied electric field.Under illumination with an intense light source, however, the response of themolecule is not any more sufficiently described by a linear effect, and nonlinearoptical (NLO) effects come into play.

Two NLO effects are of particular importance for biochemical sensing:higher harmonic generation and intensity-dependent multiphoton absorption.Higher harmonic generation involves the mixing of two or more light beamswith frequency νi to produce a light beam with frequency ν = Σνi. Multipho-ton absorption describes the simultaneous absorption of two or more photonsto reach a particular excited energy state [4].

Since the efficiency of both processes is a sensitive function of a molecule’senvironment, these NLO effects can effectively be used for biochemical sensing.However, because intense laser light is required, NLO effects are rarely usedin optical biochips.

12.4 Preferred Sensing Principles for Optical Biochips

Because of the simplicity of the practical realization and the sensitivity ofthe methods, two basic sensing principles are preferred in optical biochips.(1) Evanescent wave sensing, where an electromagnetic wave is propagatingalong an optical interface, and a part of its energy is detecting changes inthe refractive index in the sample volume. (2) Fluorescence sensing, whereincident light is exciting the fluorescence labels on the analytes whose lightemission is detected.


Fig. 12.5. Propagation of an electromagnetic wave in a dielectric slab waveguide(left) or as a surface plasmon at a metal–dielectric boundary (right)

12.4.1 Evanescent Wave Sensing

Electromagnetic waves can be guided by geometrically confining the spacewhere they should propagate. Two particular configurations are commonlyemployed in optical biochips, as illustrated in Fig. 12.5: slab waveguides, con-sisting of a stack of laterally extended thin films of dielectric materials, with acore layer whose refractive index is larger than that of the neighboring layers.Alternatively, boundary layers between a dielectric and a metallic mediumare used, where the dielectric constant of the metal must be negative in theused wavelength range. At such a boundary, a special guided electromagneticmode, a so-called surface plasmon, can propagate.

In the case of a slab waveguide, the guided modes in most waveguidetypes can be classified in TE and in TM modes, where the electrical (TE),respectively, the magnetic (TM) field has a dominant component in the lateraldirection, i.e., for the coordinate y in Fig. 12.5. Since three of the field compo-nents are zero, the TE modes consist of Ey,Hy, and Hz, and the TM modesconsist of Hy, Ex, and Ez. Additionally, it is found that the field componentsin the propagation direction,Hz, respectively, Ez, are usually quite small [2].Solving Maxwell’s equation for the slab waveguide geometry and for a givenvacuum wavelength λ of the light, one obtains for the jth TE mode an electricfield component Ej(x, y, z, t, λ) in y-direction, which is expressed as

Ej(x, y, z, t, λ) = Ej(x, y, λ) ei(neff (λ)kz−ωt), (12.1)

with the wave vector k = 2π/λ = ω/c. This equation describes a harmonicwave propagating in z-direction with the velocity v = c/neff , where the effec-tive refractive index neff is limited to the range

nc > neff > (n1, n2). (12.2)

The transverse electric field amplitude Ej(x, y, λ) is called the field profile ofthe electromagnetic mode number j, and the modes are numbered accordingto their neff values in descending order, i.e., the mode with the highest neff

corresponds to the lowest mode number j.

226 P. Seitz

The number of guided modes that can propagate in a dielectric slab wave-guide, their neff values and their field profiles Ej(x, y, λ) depend on the geom-etry of the slab waveguide and the actual distribution of refractive indices.Configurations exist in which only one TE mode (and often one TM mode) canpropagate; such structures are called monomodal or mono-mode waveguides.

If the dielectric slab waveguide is lossy, for example, through absorption orscattering, the effective refractive index neff in (12.1) has a nonzero imaginarycomponent expressing these losses. High-quality dielectric slab waveguideswith negligible losses can be fabricated, with absorption constants that aremuch lower than 1 cm−1.

The surface plasmon, the special electromagnetic mode that can propagatealong a metal–dielectric interface, is a single TM mode: At each wavelengthonly one surface plasmon can propagate, and it requires that the dielectricconstant of the metal is negative in the used wavelength range. In the visibleregion, this is true for metals such as gold, silver, and copper. The propagationvelocity in z-direction is given by v = c/neff , and the effective refractiveindex neff depends on the relative dielectric constants of the metal and of thedielectric material [4]. In contrast to the case of the dielectric slab waveguide,the propagation length of a plasmon is quite limited since the metal is stronglyabsorbing; the propagation length of a surface plasmon is usually smaller than100 µm.

The importance of the dielectric slab waveguide and of the metal–dielectricinterface for optical biosensors is due to the fact that in both structures apart of the wave energy is propagating in the covering dielectric material withrefractive index n1. The field strength in the dielectric materials with n1 andn2 is decaying exponentially as a function of the distance to the interface,with a decay length Di (i = 1, 2), for which the field strength is reduced to1/e, given as

Di =1

k√

n2eff − n2

i

. (12.3)

This exponentially decaying field is called evanescent field, and the coveringdielectric material is actually the fluidic probe volume whose change in therefractive index is measured.

Typical values of the decay length D1 in a dielectric slab waveguide usedfor optical biosensing are 0.1 µm and slightly larger, about 0.2–0.3 µm, forsurface plasmons.

Principles of Guided Wave Coupling

Since guided waves are characterized by their property that they do not nor-mally leave the waveguide structure, special means are necessary to couplethem into or out of the waveguide. The most common coupling approachesare prism coupling, grating coupling, and end-face coupling, as illustrated inFig. 12.6. Plasmons are almost exclusively coupled in and out with a prism in


Fig. 12.6. Coupling of incident light into a waveguide using a prism (left), a gratingcoupler (middle), or a focusing lens at the waveguide’s end face (right)

the so-called Kretschmann configuration [4]. The advantage is that all com-ponents are planar and easy to coat. Grating coupling is effectively used fordielectric slab waveguides, and innovative sensor configurations can be devisedwith grating couplers due to the freedom one has with selecting special grat-ing parameters [1]. End-face coupling, however, is not often employed, mostlydue to the alignment and stability problems encountered in practice.

Waveguides with Prism Coupler

Snell’s law for the transmission and reflection of light at an optical interfacestates that the product of refractive index times the sine of the propagationangle with the surface normal must be identical for incident and for trans-mitted light [16]. Since we demand from a prism coupler that light incidentunder an angle Θ is guided into a waveguide parallel to the prism’s surface,the sine of the output angle equals unity, and the coupling condition for theprism coupler simply requires that

neff(λ) = np(λ) sin θ, (12.4)

where np is the refractive index of the prism; a standard glass prism hasnp

∼= 1.5. If a prism coupler is used for an optical biosensor using surfaceplasmons, it is important that the metal film on the prism is thin enoughso that the evanescent wave on the metal’s other side is formed without toomany losses. In practice, the thickness of the metal film is typically 40–50 nm.

Waveguides with Grating Coupler

The same reasoning as for the prism coupling condition applies to the case ofa grating coupler with period Λ. The grating coupling condition is given by

neff(λ) = n2(λ) sin θ − mλ

Λ. (12.5)

According to (12.2), the effective refractive index neff of the guided mode islarger than the refractive index of the substrate n2. It is immediately obvi-ous from (12.5) that light can be coupled in from any substrate, even from

228 P. Seitz

air (n2 = 1), since the integer diffraction order m, which can also be a neg-ative number, gives us a large degree of freedom for our choice of couplingconfiguration.

In practice, coupling gratings are very shallow and the core layer of thedielectric slab waveguide is rather thin. Typical values are 100–200 nm for thecore thickness, 300–500 nm for the grating period, and 5–50 nm for the gratingdepth [1].

Waveguides with End-Face Coupler

Since the core of the waveguide is so thin, the “obvious” end-face coupling oflight with a focusing lens, as illustrated in Fig. 12.6, is very difficult to achieve.The alignment tolerances and the mechanical stability of the optical setupmust be below 100 nm, which is quite difficult to achieve in practice, whereoptical biochips should be disposable, quick to read out, and mechanicallyvery robust. Therefore, end-face coupling is not used in commercial opticalbiosensors.

Resonance Condition for Evanescent Wave Sensing

From (12.4) and (12.5) it becomes clear how measurements are carried out inoptical biosensors: The presence of the analyte subtly changes the refractiveindex in the sample volume. This change is sensed by the evanescent wave,since it implies a change in neff of the guided wave. A change in neff alters thecoupling condition, which can be adjusted either with the in- or out-couplingangle Θ or with the wavelength λ. For this reason, evanescent wave sensingin optical biochips as described above is a resonance method: The amount oflight in the guided wave is maximum for the combination of Θ and λ thatfulfills the coupling condition for the present value of neff which is dependenton the analyte concentration.

The most sensitive methods for measuring this resonance condition arecapable of resolving changes in the refractive index of about 10−7 [1, 2]. Thiscorresponds to a mass detection sensitivity of nearly 100 fg mm−2, and themolar concentration of the analyte can be determined with a sensitivity ofabout 10−11, i.e., the concentration sensitivity is 10−11 mol l−1. Note that it isimportant to stabilize for temperature variations at these high measurementsensitivities, since ∂n/∂T ≈ 10−4 K−1 for water at room temperature [17].

12.4.2 Fluorescence Sensing

Since fluorescence (or luminescence) sensing requires the imaging of two-dimensional distributions of excited or self-luminous light patterns, the usualsensing technique is fluorescence (or luminescence) microscopy [4]. For thispurpose, either conventional optical microscopes or scanning (confocal) op-tical microscopes are employed, where the sample is typically illuminated


through the same lens system through which the fluorescence light is collected(so-called epi-fluorescence excitation setup) [4].

The high demands on the collection and detection efficiency of the emittedlight in fluorescence (and luminescence) sensing could also be met with analternative, as yet little explored sensing approach. Instead of using an inactivesubstrate such as glass or plastic for the fixation of the receptor molecule, thereceptors can be bonded directly to the surface of a solid-state image sensor.In this way, a large part of the light that is emitted close to the pixel surface,usually less than 1 µm from the photosensitive semiconductor material, can becollected and detected. Although this approach has been studied by severalcompanies, no product has yet appeared on the market. This is probablydue to the current cost of high-sensitivity image sensors. Since it is expectedthat the dropping prices and the increased sensitivity of cell phone cameraswill soon have an influence on the use of these components in other areas,in particular in the life sciences, this alternate approach to fluorescence (andluminescence) sensing might become important in the future.

12.5 Readout Methods for Evanescent Wave Sensors

As mentioned earlier, the analyte concentration is measured in surface plas-mon or in dielectric waveguide sensors by determining the maximum of aresonance effect. The following realizations of optical readout principles havebeen successfully employed in practice for this purpose:

12.5.1 Angular Scanning

The most common methods to satisfy the in-coupling condition (12.4) or(12.5) is to work with a convergent beam of light or to mechanically adjustthe angle of a parallel light beam. In both cases, it is assured that a prop-agating mode is created at the sensor surface. While it is much simpler toemploy a convergent beam of light, the efficiency is much lower compared toa mechanically adjusted system because the in-coupling condition is fulfilledonly for a small fraction of the incident light.

No mechanical adjustment must be foreseen on the detection side, however,because one- or two-dimensional solid-state image sensors are employed, whichcan detect the resonance maximum without any loss of light.

In Fig. 12.7, a complete such system based on prism coupling is schemat-ically illustrated, together with typical sensor signals acquired with a solid-state line-sensor. In this example, a convergent beam of light is employed, sothat the complete system can be realized without any mechanically adjustablecomponents. Such systems are commonly used with surface plasmon sensing.

230 P. Seitz

Fig. 12.7. Schematic illustration of an optical biosensor based on prism coupling(left) and typical line-sensor signals acquired with the system (right)

12.5.2 Wavelength Tuning

The light loss or the complexity of angular adjustment of the input light can beavoided if the wavelength of the incident light is modified instead. An efficient,all solid-state optical biosensor results, in which the wavelength tuning of thelight source is realized by tuning the drive current of a VCSEL (vertical cavitysurface emitting laser diode) [18].

12.5.3 Grating Coupler Chirping

Another approach to avoid the mechanical adjustment of the angles of theincident and the out-coupled light is to work with gratings whose period isvaried laterally. The use of such chirped gratings ensures that there is alwaysa part of the grating for which the in- or out-coupling condition (12.5) is ful-filled. This technique has the interesting property that the lateral propagationlocation of the guided wave is a direct measure of the resonance position. Forthis reason, the method has also been called the “light pointer” approach tooptical biosensing [19].

A significant disadvantage of the chirped grating technique is the low effi-ciency with which the available light is used, similar to the convergent beamapproach described in Sect. 12.5.1. Only the fraction of light for which the in-coupling condition is fulfilled can propagate in the waveguide and contributesto the sensing signal; the largest part of the optical energy is just transmittedthrough the grating.

12.6 Substrates for Optical Biochips

Although optical biochips are most often disposable devices, the optical sens-ing techniques employed put high demands on the surface quality, the mechan-ical stability, and the robustness of the used materials. At the same time, thematerials should be of low cost, easy to process, and of low environmental con-cern. For these reasons, the dominant substrate materials for optical biochips


are glass, fused silica (high purity synthetic amorphous silicon dioxide), ormono-crystalline silicon as used in large quantities for the semiconductor in-dustry. Since the interface between the coupling prism and the biochip doesnot need to be patterned – it must only be optically flat – glass or fused silicaare the preferred substrate materials for the prism coupling approach. This isparticularly true for surface plasmon resonance (SPR) biosensors.

On the other hand, grating coupling requires the fabrication of couplinggratings on the substrate. Although glass is still the material of choice for thisapplication, the use of plastic or polymeric materials such as PMMA (poly-methylmethacrylate), PDMS (polydimethylsiloxane) or OrmoceresTM is beinginvestigated, since surface gratings can easily be replicated in these materialswith nanometer precision. Before these materials are employed routinely inpractice, however, several problems regarding their surface roughness, the me-chanical stability, and the physical/chemical inertness of the surfaces need tobe resolved satisfactorily. An additional problem that needs to be addressedis the thermal expansion coefficient of polymeric materials, which is typicallyten times as large as the one of glass or silicon [17].

Fluorescence and luminescence sensing, in particular, using the specialtypes of microarrays discussed in Sect. 12.2, is most often realized on glassor silicon substrates. Since the materials must neither be transparent norof particular optical quality, the choice of cost-effective substrates is largerthan that for the other optical biosensors. Because of the ease with which themicrofluidic functionality can be integrated into plastic or polymeric materials(e.g., by injection molding), these materials are also of interest for fluorescencebiosensors [20].

12.7 Realization Example of an OpticalBiosensor/Biochip: WIOS

As a practical realization example of an optical biochip and its associatedreadout system, the wavelength-interrogated optical sensor (“WIOS”) princi-ple described in [18] is explained in more detail. This label-free optical biosen-sor is based on a dielectric waveguide with grating couplers for coupling lightinto and out of the waveguide.

The disposable biochip consists of a 12.5×12.5 mm2 AF45 glass substrateof 0.7 mm thickness covered with a 150 nm thick film of Ta2O5 (n = 2.13)with a grating structure that was previously etched into the glass using dryetching. This grating has a period of Λ = 360 nm and a thickness of 13.2 nm.To minimize interference effects between the input and the output grating,the Ta2O5 film thickness was fabricated 150 nm thicker at the output grating,resulting in a smaller out-coupling angle. Since the complete input/outputgrating structure measures only 0.8×1.0 mm2, several of them can be producedon the same biochip. The input grating of the sensing pads is selectivelysensitized with suitable receptor molecules that are photo-bonded to the input

232 P. Seitz

Fig. 12.8. Measurement principle of the WIOS optical biosensor [18]

grating surface using OptoDexTM photolinker material. The sensor chip isthen placed into a fluidic cell and sealed.

The disposable biochip is read out with a compact optical system illus-trated schematically in Fig. 12.8. It consists of a VCSEL light source with cen-ter wavelength of 763 nm and an electrical wavelength-tuning range of 2 nm.The laser beam is collimated and expanded to cover the input pad of thebiosensor with a suitable optical sub-system. Because of the limited wave-length tuning range, an additional deflection mirror has been placed into theinput beam path, so that a deflection range of 6◦ is obtained. The light fromthe output grating is collected by an optical multi-mode fiber and transportedto a silicon photosensor array for detection, amplification, and analysis. In op-eration, the VCSEL current is modulated with a sawtooth function, resultingin a sawtooth modulation of the VCSEL’s emission wavelength. The photode-tector’s signal is analyzed for the precise temporal position of the resonancepeak, and this time shift is a direct measure for the refractive index changeand the analyte concentration in the sample volume. As described in [18], theWIOS biochip and readout instrument reach an experimental uncertainty ofthe refractive index measurement of 3.1× 10−7, corresponding to a mass cov-erage detection standard deviation of about 100 fg mm−2. These values wereobtained for the rather small Biotin molecule with a molecular weight of only244 Da, using neutravidin as the receptor.

12.8 Outlook: Lab-on-a-Chip Using OrganicSemiconductors

In the future, optical biosensors will contain more and more functionality onthe same substrate for increased ease-of-use, higher reliability, and lower cost.This will lead to highly integrated lab-on-a-chip systems that will see a large


number of additional uses at home, in industry, in safety/security, and forenvironmental applications. Current approaches employ many different mate-rials to perform the various required functions: Optical, mechanical, fluidic,optoelectronic (light generation and detection), signal processing, informationdisplay, etc. The resulting solutions are hybrid systems, consisting of manydifferent components, whose integration into a final product is not very cost-effective.

There is a promising class of materials, however, that might make it pos-sible to realize a monolithic lab-on-a-chip for low-cost optical biosensors: or-ganic semiconductors, in particular, polymer semiconductors. In the following,a few salient points of this material class and its potential for the fabricationof highly functional yet cost-effective optical biosensors are summarized.

12.8.1 Basics of Organic Semiconductors

The scientific field of organic semiconductors is barely 30 years old, and itstarted with the development of organic light-emitting diodes (LED) withuseful conversion efficiency by the American company Kodak. Since then,this field has progressed very rapidly, due to the promise of a new class ofmaterials offering all the optoelectronic functionality required for the realiza-tion of generic photonic microsystems: light generation, light detection, analogand digital electronic signal processing, as well as photovoltaic power gener-ation [21]. In addition, efficient and fast production methods are available,which are often modified printing techniques such as inkjet, silk-screen, orgravure printing. Therefore, it is anticipated that the fabrication costs willbe about 100 times lower than that for silicon-based semiconductor circuitsin the near future (¤0.1 cm−2 compared to ¤10 cm−2 for standard CMOSchips).

However, the performance of organic semiconductors can be significantlylower in several respects than the performance of their inorganic counterparts.The charge carrier mobility is about 1,000 times lower (about 1 cm2 V−1 s−1

compared to 103 cm2 V−1 s−1), the diffusion length of good material is severalorders of magnitude smaller (typically 10 nm compared to several 100 µm), theelectrical resistivity is significantly higher (since it is much more difficult todope organic semiconductors), and organic semiconductors deteriorate muchmore rapidly if exposed to ultraviolet light, to oxygen or to water vapor.

For these reasons, organic semiconductors are not a cheaper and simplerreplacement of inorganic semiconductors, and it is rather necessary to evaluatethe performance of each device individually for the intended application.

12.8.2 Organic LEDs

Since a large variety of organic semiconductors are available, organic LEDs(oLED) of many different colors can be fabricated [22]. Their external quan-tum efficiency can exceed 10%, and their power efficiency already surpasses

234 P. Seitz

the performance of conventional light sources. The highest luminous efficacyof a white oLED reported in mid 2006 by the Japanese company KonicaMinolta is 64 lm W−1, which compares very favorably with the luminous effi-cacy of conventional light bulbs (15 lm W−1), halogen lamps (25 lm W−1), orcompact fluorescent lamps (60 lm W−1).

Because of the simplicity and the efficiency with which oLEDs – also ofdifferent color – can be produced on large surfaces, they are increasingly usedfor the fabrication of displays in cameras, for PCs, and for TV screens.

12.8.3 Organic Lasers

The first optically pumped organic (polymer) laser has been demonstratedalready in 1996 [23]. However, because of the high electrical resistivity oforganic semiconductors, the electrically pumped organic CW laser working atroom temperature remains still elusive.

The fact that no intense, narrow-band or even monochromatic light sourcescan currently be fabricated with organic semiconductors is one of the majorshortcomings for the realization of monolithic optical biochips and labs-on-a-chip using solely organic materials.

12.8.4 Organic Photodetectors and Image Sensors

Once the problem of dissociation of photogenerated, tightly bound chargepairs (Frenkel excitons) had been solved using finely dispersed polymer blendheterojunctions, the quantum efficiency of photodetectors fabricated with or-ganic semiconductors approached that of inorganic semiconductors [24]. Totalexternal quantum efficiencies exceeding 50% are routinely obtained today inphotosensors realized with organic semiconductors. By using thin multilayerstructures, it is also possible to produce high-speed photodetectors despite thelow mobility of charge carriers in organic semiconductors: Maximum detectionfrequencies of about 500 MHz have already been demonstrated.

A particular advantage of photodetectors fabricated with organic semicon-ductors is the high number of sensors (pixels) that can be realized simultane-ously on a very large substrate. As an example, the cost-effective fabricationof page-sized image sensors has already been demonstrated.

Since different organic semiconductors with various spectral responses canbe easily integrated on the same substrate, the fabrication of color photo-sensors with good colorimetric performance has been achieved already earlyin [24].

12.8.5 Organic Photovoltaic Cells

One of the major problems of a practical, disposable optical biochip is electri-cal power supply. If it were possible to replace the conventional electrochemical


batteries of environmental concern with another, less harmful source of electri-cal power, this would add considerably to the attractiveness of an all-organicoptical biochip solution.

Since it is possible to fabricate efficient photodetectors using organic semi-conductors, one immediately thinks of the monolithic cointegration of photo-voltaic cells on an optical biochip. This is indeed feasible, and all-organic solarcells have been successfully demonstrated. Because of the relatively large frac-tion of power contained in the near infrared spectral range of sunlight, whereorganic semiconductors are notoriously insensitive, the total quantum effi-ciency of organic photovoltaic cells is rather limited. The maximum quantumefficiency demonstrated up to date does not exceed 5%.

12.8.6 Organic Field Effect Transistors and Circuits

The first field-effect transistors (FETs) and electronic circuits based on or-ganic semiconductors were demonstrated in 1995 [25]. The electrical charac-teristics of these organic FETs are similar, in principle, to the one of inorganicFETs. However, because of the small charge carrier mobility and low electri-cal conductivity of organic semiconductors, the speed and the current densityof organic FETs are much lower than those in their inorganic counterparts.The switching speed is currently limited to a few megahertz, and the high-est frequency at which an organic circuit has worked to date is 13.56 MHz, afrequency of importance for RF-ID electronic tags and inductive data com-munications.

An additional problem is the difficulty of cointegrating both n-MOS andp-MOS transistors with comparable performance using compatible organicsemiconductors. For this reason, power-efficient CMOS circuits cannot yet beroutinely integrated with organic semiconductors, and ultra-low-power elec-tronics still remains in the realm of inorganic semiconductors.

12.8.7 Monolithic Photonic Microsystems Using OrganicSemiconductors

Organic semiconductors allow the fabrication of similar electronic and opto-electronic components as produced with inorganic semiconductors. In almostno respect are organic components superior in their performance comparedto their inorganic counterparts. The huge appeal of organic semiconductors isin the ease, rapidity, and low cost of their production using modified printingtechniques on large areas (several square meters) with very little material: only1 g of organic semiconductor is sufficient to produce about 10 m2 of “plasticoptoelectronics.” Another unique property offered by organic semiconductorsis the possibility of monolithic fabrication of complete photonic microsystemson a single substrate: Analog and digital electronics, light generation, lightdetection, and even a photovoltaic power supply can all be integrated on the

236 P. Seitz

same chip [26]. Once the diffusion barrier and sealing problem is solved, itwill even become possible to produce these complete photonic microsystemson flexible substrates with a lifetime of many years.

12.9 Conclusions and Summary

Optical biochips of various integration levels have become important toolsin many different areas of the life sciences: Genomics, proteomics, medicaldiagnostics, pharmaceutical drug screening, contamination detection in thefood industry, environmental and pollution monitoring, as well as counter-terrorism all make substantial use of this technology, due to its simplicity,reliability, sensitivity, and low cost. Increasing levels of integration will soonlead to complete, disposable labs-on-a-chip with even higher functionality andperformance. The availability of organic semiconductors makes it possible tofabricate complete monolithic photonic microsystems on a single substrate,for example, on an injection-molded piece of polymer containing the entiremicrofluidics part of a lab-on-a-chip. This will open up many more appli-cation areas to optical biochips, and it will make them highly versatile andindispensable tools of our everyday lives.

Acknowledgments

This contribution could not have been prepared without the invaluable helpof M. Wiki and R.E. Kunz, both at CSEM SA, whose generous support Igratefully acknowledge.

References

1. R.E. Kunz, in Integrated Optical Circuits and Components, ed. by E.J. Murphy(Marcel Dekker, New York, 1999), p. 335

2. P.V. Lambeck, Meas. Sci. Technol. 17, 93 (2006)3. J. Tschmelak et al., Talanta, 65, 313 (2005)4. P.N. Prasad, Introduction to Biophotonics (Wiley, Hoboken, NJ, 2003)5. Affymetrix Inc., 3380 Central Expressway, Santa Clara, CA 95051, USA,

http://www.affymetrix.com6. M. Schena (ed.), Microarray Biochip Technology (Eaton Publishing, Natick,

MA, 2000)7. E. Oosterbroek, A. van den Berg (eds.), Lab-on-a-Chip (Elsevier, Amsterdam,

Netherlands, 2003)8. A. Manz, N. Graber, H.M. Widmer, Sens. Actuators B 1, 244 (1990)9. E.J. Cho, F.V. Bright, Anal. Chem. 73, 3289 (2001)

10. A.C.R. Grayson et al., Proc. IEEE 92, 6 (2004)11. R. Bahir, Adv. Drug Deliv. Rev. 56, 1565 (2004)


12. H.J. Fecht et al., Nanotechnology Market and Company Report 2003 (WMtechand University of Ulm Publishing, D-Ulm, 2003)

13. J. Ziauddin, D.M. Sabatini, Nature 411, 107 (2001)14. J.A.B. Kononen et al., Nat. Med. 4, 844 (1998)15. P. Seitz, in Computer Vision and Applications – A Guide for Students and

Practicioners (Academic Press, San Diego, CA, 2000)16. J.D. Jackson, Classical Electrodynamics, 3rd edn. (Wiley, New York, 1998)17. D.R. Lide (ed.), CRC Handbook of Chemistry and Physics, 87th edn. (Taylor &

Francis, Bota Raton, FL, 2006)18. K. Cottier et al., Sens. Actuators B 91, 241 (2003)19. R.E. Kunz et al., Sens. Actuators A 47, 482 (1995)20. F. Fixe et al., Nucleic Acids Res. 32, e9 (2004)21. S.R. Forrest, IEEE J. Sel. Top. Quantum Electron. 6, 1072 (2000)22. R.H. Friend et al., Nature 397, 121 (1999)23. M.D. McGehee, A.J. Heeger, Adv. Mater. 12, 1655 (2000)24. G. Yu et al., Synth. Met. 111–112, 133 (2000)25. E. Cantatore, in Proceedings of ESSDERC 2001, (Nurnberg, Germany, 11–13

Sept. 2001)26. P. Seitz, U.S. Patent 7,038,235, 2 May 2006

13

CMOS Single-Photon Systems for BioimagingApplications

E. Charbon

13.1 Introduction

Recent advances in neurobiology and medical imaging have put an increasingburden on conventional sensor technology. This trend is especially pronouncedin time-correlated imaging and other high precision techniques, where timingaccuracy and sensitivity is critical.

Techniques have been proposed to achieve high speed in conventionalcharge-coupled devices (CCDs) and CMOS active pixel sensor (APS) archi-tectures [1–4]. Among some of the most successful techniques are ultra-fastlow-noise electronic readout circuitries, on-pixel A/D conversion, and localanalog electrical storage. However, in general a significant design effort andconsiderable experience is needed to achieve satisfactory results.

Alternatives to conventional CCDs and CMOS APS sensors are single pho-ton counters (SPCs). Several types of SPCs have been known for decades.Among the most successful devices in this class are microchannel plates(MCPs) and photomultiplier tubes (PMTs) that have become the sensorsof choice in many applications [5]. Even though they have been studied sincethe 1960s [6], silicon avalanche photodiodes (SiAPDs) have become a seriouscompetitor to MCPs and PMTs only recently. In SiAPDs, carriers generatedby the absorption of a photon in the p-n junction are multiplied by impactionization; thus producing an avalanche. The resulting optical gain is usuallyin the hundreds. The main drawback of these devices, however, is a relativelycomplex amplification scheme and/or complex ancillary electronics. In addi-tion, specific technologies are often required.

If biased above breakdown, a p-n junction can operate in so-called Geigermode. Such a device is known as single photon avalanche diode (SPAD). InGeiger mode of operation, SPADs exhibit a virtually infinite optical gain;however, a mechanism must be provided to quench the avalanche. There ex-ist several techniques to accomplish quenching, classified in active and pas-sive quenching. The simplest approach is the use of a ballast resistance. The

240 E. Charbon

avalanche current causes the diode reverse bias voltage to drop below break-down; thus pushing the junction to linear avalanching and even pure accumu-lation mode. After quenching, the device requires a certain recovery time toreturn to the initial state. The quenching and recovery times are collectivelyknown as dead time.

Recently, SPADs have been integrated in CMOS achieving timing res-olutions comparable to those of PMTs [7]. Current developments in moreadvanced CMOS technologies have demonstrated full scalability of SPADdevices, a 25 µm pitch, and dead time as low as 32 ns. The sensitivity, char-acterized in SPADs as photon detection probability (PDP), can exceed 25–50%. The noise, measured in SPADs as dark count rate (DCR), can be aslow as a few hertz [8, 9]. Thanks to these properties, SPCs based on SPADshave been proposed for imaging where speed and/or event timing accuracyare critical. Such applications range from fluorescence-based imaging, such asForster Resonance Energy Transfer (FRET), fluorescence lifetime imaging mi-croscopy (FLIM) [10], and fluorescence correlation spectroscopy (FCS) [11], tovoltage sensitive dye (VSD) based imaging [12,13], particle image velocimetry(PIV) [14], instantaneous gas imaging [15,16], etc.

In the following sections we explore some applications of SPCs in com-parison to conventional sensors, including potential fields of imaging whereSPADs can be a compelling implementation aspect of SPCs. We also lookat performance and architectural issues that the designer needs to take intoaccount when approaching real problems involving the use of SPCs.

13.2 Spectroscopy

Spectroscopy-based imaging has many incarnations. One, very successful one,is known as fluorescence correlation spectroscopy (FCS); a technique used tomeasure transitional diffusion coefficients of macromolecules to count fluores-cent transient molecules or to determine the molecular composition of a fluidbeing forced through a bottleneck or a gap. In FCS, a femtoliter volume isexposed to a highly focused laser beam that causes the molecules in it toemit light in a well-defined spectrum. Figure 13.1 shows an example of opticalmolecular response depending on the size and diffusion pattern of the mole-cule. The physical causes of this behavior are related to the mobility of theligands. In the first case, free fluorescent ligands are continuously entering andleaving the detection volume. In the second case, macromolecule ligands areless mobile; thus producing slower but highly correlated intensity fluctuations.Figure 13.2 shows an example of typical autocorrelation functions simulatedfor different molecules [17].

A tighter correlation is observable in the case of low molecular weightligands. A macromolecule ligand generates a much more relaxed correlation.A mixture of free and bound ligands is shown in the middle curve.

13 CMOS Single-Photon Systems for Bioimaging Applications 241

Time Time

Flu

ores

cenc

e

Flu

ores

cenc

e

Fig. 13.1. Optical response of molecules bombarded by highly focused laser beam.Rapidly diffusing small molecule (left); slow, large molecule, with its large well-defined bursts of optical energy (right)

Fig. 13.2. Autocorrelation of fluorescence response. High time resolution is espe-cially useful for sub-nanosecond response dynamics

Dual-color cross-correlation FCS measures the cross-correlation of the flu-orescence intensities of two distinct dyes [18]. Thus, it becomes possible todetect different molecules without reference to their diffusion characteristics.More recently, a combination of the two methods has been reported to re-duce the need for a distinct, often rather large mass ratio between the twomolecule types [19]. A typical FCS setup is shown in Fig. 13.3. A highly fo-cused laser beam is directed towards the detection volume. The reflected lightis applied to the high sensitivity optical sensor through a prism. A digitalcorrelator matches the autocorrelation function with a database of known re-sponses. Generally, in FCS time resolutions of a few tens of picoseconds andsensitivities equivalent to a few hundred photons are needed.

242 E. Charbon

Fig. 13.3. Basic FCS setup (gap volume not to scale). Single or multipixel detectorscan be used in this setup. With multipixel imaging devices information can be gainedfrom the emissions from the surroundings of the molecule

13.3 Lifetime Imaging

Among time-correlated imaging methods, time-correlated single photoncounting (TCSPC) is perhaps one of the most used in bioimaging. Mul-tiple exposures are employed to reconstruct the statistical response of matterto sharp and powerful light pulses. The statistics are generally represented inform of a histogram, while light source repetition frequencies may vary fromkilohertz to hundreds of megahertz. From the response statistics, biologistsgenerally extract parameters that can be used to characterize the moleculeunder observation and/or its environment, e.g., the calcium concentration.

The study of calcium at the cellular level has made intensive use of flu-orescent Ca2+ indicator dyes. Examples of heavily used dyes or fluorophoresare Oregon Green Bapta-1 (OGB-1), Green Fluorescent Protein (GFP), andmany others. Calcium concentration can be determined precisely by measur-ing the lifetime of the response of the corresponding fluorophore, when excitedat a given wavelength.

Using FLIM, for example, one can determine the variation of calcium con-centration in neural cells as a function of a given activity. Figure 13.4 showsa conceptual setup where a neural cell soma is being exposed to light via,for example, a fiber. The fluorophore molecules, previously injected into thecell, exhibit different lifetimes depending upon the calcium concentration inthe vicinity of ion channels. There exist several flavors of FLIM based onhow lifetime is characterized or based on the excitation mode. In one-photonFLIM, for example, only one photon is required to force a state change inthe fluorophore molecule [10]. In this case, only a small shift in wavelengthis observed between the excitation and response. As a consequence, filteringthe excitation pulse from the measurement response may be challenging. Inaddition, any scattering occurring along the observation path induces pho-tonic noise (usually at the same or close wavelength of fluorescence emissions)


Fig. 13.4. Calcium environment in and around a cell membrane. Calcium movesthrough the membrane via so-called ion channels. A fluorophore can detect smallvariations of calcium concentration

into the measurement. Nonetheless, the electro-optical setup required by thismethod is generally less critical and relatively easy to build.

Multiphoton FLIM, even though conceptually known for decades, only re-cently has become an essential tool for neurobiology and other disciplines.Two-photon FLIM has been the method of choice, thanks to the recent ad-vances in laser technology that can now concentrate kilowatts of light powerin micrometric volumes of matter. There are several advantages to two-photon FLIM. First, fundamental spatial confinement for the excitation canbe achieved; thus allowing one to isolate a single molecule or cluster of mole-cules. Second, because of the reduced average optical powers in play, effectssuch as photobleaching and photoxicity can be mitigated; thereby enhancingthe suitability of the approach. Third, a better penetration in turbid mediacan be achieved, due to reduced scattering. Fourth, the large spectral differ-ence between in- and out-going radiation simplifies the separation of responsefrom excitation [10].

The main components of typical two-photon FLIM setups are a high-powerfemtosecond source, generally a mode-locked Ti:Sapphire laser; a SPC; timediscrimination hardware; a standard laser-scanning microscope. An exampleof laser-scanner based FLIM image is shown in Fig. 13.5. In this experimentit was possible to detect intra-cellular chemical waves to help identify inner-workings of pathogens or the impact of certain pharmaceuticals [20]. Thebinding of neutrophil at the interface of two endothelial cells was monitoredat high speed via calcium-triggered chemical waves propagating through thecells.

244 E. Charbon

Fig. 13.5. High-speed image sequence of the binding of neutrophil. (Courtesy ofH.R. Petty [20])

Fig. 13.6. Coplanar gamma emission where a positron annihilates with an electron.The synchronicity of the emission may be used to infer the loci of the emission usingmultiple synchronized detectors

13.4 Time-of-Flight in Bio- and Medical Imaging

Time-of-flight (TOF) is the time a light ray takes to propagate between twopoints in the three-dimensional space. There exist several applications requir-ing a precise measurement of TOF to image particular properties of targetsand environments. Let us consider two examples of such applications: positronemission tomography (PET) and depth map imaging.

In a PET system, a positron is emitted in the tissue being imaged by avariety of substances. An example of one such substance is fluorodeoxyglucose(FDG). This compound is absorbed by human cells, in the brain for example,at different rates based on the operation – whether normal or abnormal – ofthe cell. After positron emission, a annihilation occurs followed by gamma rayemission.

Gamma emission is well defined spatially and temporally. In fact, anni-hilation results in a pair of gamma photons in opposite directions (coplanarproperty) at exactly the same time. Figure 13.6 shows schematically howemission occurs. To detect gamma emission one usually utilizes a crystal, e.g.,lutetium oxy-orthosilicate (LSO), which converts gamma into visible radia-tion. Typically several thousands photons may be released in this fashion.The timing resolution of detection via LSO, however, is generally worseneddue to the properties of these crystals. The lifetime of a typical LSO crystalmay be as much as 39 ns; hence statistics must be used in combination withdetector with high timing resolution to derive the first non-noise photon tobe detected (Fig. 13.7). Such photon will give an approximation to the actualmoment in time when the gamma radiation is released.


Fig. 13.7. Typical crystal photoemission probability as a function of time sincegamma absorption

Fig. 13.8. Result of time correlation that enables the detection of the source ofgamma radiation

To find the exact location of positron emission, one must monitor allgamma radiation reaching a pair of detectors on an axis at exactly the sametime and then cross-correlate all estimated arrival times. The emission locimay be derived by measuring the TOF of the particle with respect to a refer-ence point of known coordinates. The tomography of emissions may be con-structed using conventional Fourier transform techniques in combination witha mechanical system where a detector pair rotates around an axis longitudinalto the cylindrical volume being probed (Fig. 13.8). The pair may be furthertranslated after each rotation cycle to complete a cylindrical scan. The litera-ture on TOF imaging and gated sensors (both CCD and CMOS APS) is veryextensive, a few examples are found in [8, 21–23].

13.5 System Considerations

A SPC may be implemented in a number of ways. If the application requiresan array of simultaneously operating single photon detectors (SPDs), thena SPAD array is a desirable alternative to PMT or SiAPD arrays in termsof cost, power consumption, and miniaturization. Several demonstrations ofCMOS SPAD arrays exist in various technologies [8, 9, 24, 25]. The methods

246 E. Charbon

Table 13.1. Performance parameters of a generic time discriminator implementedin a conventional IC technology

Measurement Value Unit

Timing resolution 1 ∼ 100 psTiming accuracy < 30 psTemperature stability 10 ∼ 1, 000 ppm/◦CDead time 1 ∼ 1, 000 ns

to read the output of every SPD pixel to the external world range from pixelrandom access, similar to APS architectures, to event-driven approaches [9,25]to pipelined readout methods [26].

Time-correlated counting can be performed either via a fast clock or dedi-cated time discrimination circuit (Table 13.1). Time-uncorrelated counting onthe contrary requires only a counter whose bandwidth is determined by theinverse of the dead time of the measurement. If counting is performed on-pixel,the operation of the sensor becomes relatively straightforward, but the sili-con real estate may not be utilized efficiently. In principle, this configurationenables the best time utilization since it maximizes parallelism.

If counting is performed on-column, SPDs may be implemented in asmaller area and a smaller pitch may be obtained; however, a mechanismmust be devised for pixels to share the column counter. As a consequence, thetime utilization efficiency is reduced and the readout complexity increases. Agood trade-off has been shown to be an event-driven readout system, whichworks best with low photon fluxes [9, 25]. In this approach a SPD uses thecolumn as a bus, accessing it only when it absorbs a photon. The saturationof the device is thus determined by the bandwidth of the column divided bythe number of rows.

A chip-wide counting is the technique that allows the simplest architecture,while it is the least time-efficient, since only one SPD may be counting at anypoint in time. Hence the maximum achievable frame rate is limited by theminimum exposure time per pixel, the total number of pixels, and the speedof the switching electronics.

Time-correlated counting may be implemented using fast time discrimina-tion circuits running at lower speed, such as time-to-digital converters (TDCs),time-to-amplitude-converters (TACs), or correlators. A TAC is a device thatconverts a time interval into a voltage difference. A TDC converts a timeinterval directly into a digital code. A correlator relates a given impulse toanother time-wise, determining in effect their phase difference. There exista wide variety of implementations for time discriminators, differentiated interms of performance, size, cost, and power dissipation. Figure 13.9 showssome of the main parameters. In the time-correlated mode of operation, alaser trigger may be used as a reference signal, call it START. The output ofa SPD may be used to terminate the time measurement, call that STOP. Themain drawback of these devices is their complexity, often requiring hundredsof transistors.


Fig. 13.9. Detector pair rotating around an axis and longitudinally translatingthrough it

On-column and on-chip approaches are, seemingly, the sole option, alongwith off-chip solutions. An example of an off-chip approach was proposedin [26]. In this design, two technologies are used to independently implementSPADs and time discriminators that are subsequently connected electricallyusing specific techniques.

The development of architectures that support time-correlated modes withsome degree of resource sharing is currently underway in many researchgroups. The main trade-off is at the architectural level, due to the natureof the signal generated by SPDs. In its most general implementation, an SPDgenerates a digital pulse when it detects a photon.

Application-specific optimal architectures are possible, provided a modelof the application is built to characterize the performance of the sensor apriori. The sharing of resources may involve a number of pixels, say 4 or 16,or on-demand sharing based upon the reaction of SPDs may be used. Othertrade-offs may include the complexity of the time discriminator itself.

13.6 Conclusions

With the introduction of CMOS single-photon avalanche diodes, it is pos-sible today to achieve great levels of miniaturization. Not only large arraysof photon counters are now possible, but also very high dynamic range andtiming accuracy have become feasible. Thanks to these advances, applicationsrequiring time-resolved single photon detection have become possible. Otherapplications have reached unprecedented levels of accuracy. We have outlinedsome of these applications and we have discussed system issues related tothese and novel applications in the field of bio- and medical imaging.

248 E. Charbon

Acknowledgments

The author is grateful to his graduate students and to David Stoppa of FBK.This research was supported by a grant of the Swiss National Science Foun-dation and the Center for Integrated Systems, Lausanne.

References

1. T.G. Etoh, Proc. SPIE 3173, 57 (1997)2. T.G. Etoh et al., IEEE Trans. Electron Devices 50, 144 (2003)3. S. Kleinfelder, S. Lim, X. Liu, A. El Gamal, J. Solid-State Circuits, 36, 2049

(2001)4. G. Patounakis, K. Shepard, R. Levicky, in IEEE Symposium on VLSI Circuits,

Kyoto, Japan, June 2005, p. 685. J. McPhate, J. Vallerga, A. Tremsin, O. Siegmund, B. Mikulec, A. Clark, Proc.

SPIE 5881, 88 (2004)6. R.H. Haitz, Solid-State Electron. 8, 417 (1965)7. A. Rochas, Dissertation, EPF-Lausanne, 20038. C. Niclass, A. Rochas, P.A. Besse, E. Charbon, IEEE J. Solid-State Circuits,

40, 1847 (2005)9. C. Niclass, M. Sergio, E. Charbon, Design and Test in Europe (DATE), Munich,

Germany, March 200610. A.V. Agronskaia, L. Tertoolen, H.C. Gerritsen, J. Biomed. Opt. 9, 1230 (2004)11. P. Schwille, U. Haupts, S. Maiti, W.W. Webb, Biophys. J. 77, 2251 (1999)12. A. Grinvald et al., in Modern Techniques in Neuroscience Research, ed. by

U. Windhorst, H. Johansson (Springer, Berlin Heidelberg New York, 2001)13. J. Fisher, Opt. Lett. 29, 71 (2004)14. S. Eisenberg et al., Proc. SPIE 4948, 671 (2002)15. S.V. Tipinis et al., IEEE Trans. Nucl. Sci. 49, 2415 (2002)16. W. Reckers et al., in Proceedings of 5th International Symposium on Internal

Combustion Diagnostics, Baden-Baden, June 200217. K.J. Moore, S. Turconi, S. Ashman, M. Ruediger, U. Haupts, V. Emerick,

A.J. Pope, J. Biomol. Screen. 4, 335 (1999)18. P. Schwille, F.J. Meyer-Almes, R. Rigler, Biophys. J. 72, 1878 (1997)19. K.G. Heinze, A. Koltermann, P. Schwille, Proc. Natl. Acad. Sci. USA 97, 10377

(2000)20. H.R. Petty, Optics Photonics News 15, 40 (2004)21. C. Bamji, E. Charbon, U.S. Patent 6,515,740, 200322. R. Lange, Dissertation, University of Siegen, 200023. R. Jeremias, W. Brockherde, G. Doemens, B. Hosticka, L. Listl, P. Mengel,

Proceedings of IEEE ISSCC, San Francisco, CA, USA 200124. D. Mosconi, D. Stoppa, L. Pacheri, L. Gonzo, A. Simoni, Proceedings of ESS-

CIRC, Montreux, Switzerland, Sept. 200625. C. Niclass, M. Sergio, E. Charbon, Proceedings of ESSCIRC, Montreux,

Switzerland, Sept. 200626. M. Sergio, C. Niclass, E. Charbon, Proc. ISSCC, p. 120, 200727. B. Aull et al., Proc. ISSCC, p. 1179, 2006

14

Optical Trapping and Manipulationfor Biomedical Applications

A. Chiou, M.-T. Wei, Y.-Q. Chen, T.-Y. Tseng, S.-L. Liu, A. Karmenyan,and C.-H. Lin

14.1 Introduction

The field of optical trapping and manipulation was started in 1970 by Ashkinet al. [1, 2], when the group at Bell Laboratory demonstrated the following:

• A mildly focused laser beam in a sample chamber containing micron-sizepolystyrene beads suspended in water could attract the bead transverselytowards the beam axis and drive the bead to move along the beam axis inthe direction of beam propagation (Fig. 14.1a).

• A pair of mildly focused laser beams aligned coaxially and propagating inopposite direction with approximately equal optical power could stably trapa bead (suspended in water) in a three-dimensional space (Fig. 14.1b).

• A mildly focused laser beam directed vertically upward can stably levitatea bead by balancing the radiation pressure against the weight of the beadin water (Fig. 14.1c).

The tremendous growth of the field was further stimulated by another mon-umental paper by Ashkin et al. in 1986 [3], when the group reported thata single laser beam alone when strongly focused with high numerical aper-ture (NA > 0.6) could stably trap micron-sized polystyrene bead in water(Fig. 14.1d). This strongly focused single-beam configuration, known as single-beam gradient-force optical trap or optical tweezers, has since become one ofthe most widely used optical trapping configurations mainly due to its relativesimplicity in experimental setup. Furthermore, potential biomedical applica-tions were soon envisioned when noncontact and noninvasive optical trappingand manipulation of biological samples such as living cells, cell organelles,and bacteria by optical tweezers with a near infrared (NIR) laser beam weresuccessfully demonstrated [4]. In addition to optical trapping and manipu-lation of a single particle outlined above, simultaneous optical trapping ofmultiple particles as well as independent manipulation of each particle hasalso been demonstrated by several methods, including, interferometric optical

250 A. Chiou et al.

(a) (b)

(c) (d)

Fig. 14.1. A schematic illustration of optical trapping and manipulation of a mi-croparticle suspended in water. (a) Transverse confinement and axial driving by amildly focused laser beam; (b) three-dimensional trapping in a counter-propagatingdual-beam trap; (c) optical levitation by a mildly focused laser beam pointing up-ward balancing against the weight of the particle; (d) three-dimensional trappingby a strongly focused (numerical aperture NA > 0.6) laser beam

patterning [5], interferometric optical tweezers [6], time-multiplexing with ascanning laser beam [7], and holographic optical tweezers [8–10]. In holo-graphic optical tweezers, a set of predesigned computer generated holograms(CGH) in a spatial light modulator is often used to create dynamic multiplefocal spots in a three-dimensional space to trap several particles and ma-nipulate each independently. Likewise, a method known as generalized phasecontrast (GPC) has been developed by Gluckstad et al. [11,12] for the manip-ulation of multiple particles. Several very entertaining and impressive videodemonstrations of multiple particle trapping and manipulation can be viewedat the website of Prof. Miles Padgett at the University of Glasgow in UK(http://www.physics.gla.ac.uk/Optics/).

Since the demonstration of optical tweezers (or single-beam gradient-forceoptical trap), the counter-propagating dual-beam trap has received much lessattention until the demonstrations that (1) the counter-propagating dual-beam trap can be implemented fairly simply, without any focusing lens,by aligning a pair of single-mode fibers face-to-face in close proximity (ap-proximately tens of microns to a few hundred microns) inside a sample

14 Optical Trapping and Manipulation for Biomedical Applications 251

chamber containing microparticles suspended in water [13,14], (2) the counter-propagating dual-beam trap can not only trap a red blood cell (RBC), osmot-ically swollen into spherical shape, but also stretch the spherical RBC todeform into an ellipsoid [15–17]. This technique has since been investigatedfor the measurement of the visco-elastic properties of various cells to correlatewith their physiological conditions, including the feasibility of identifying acancer cell against a normal cell [18].

Although a self-aligned counter-propagating dual-beam trap can be imple-mented, in principle, with a single input beam in conjunction with a nonlinearoptical phase-conjugate mirror [19], it has not been further developed due tosome practical limitation in the nonlinear optical properties of the materi-als available for optical phase-conjugation. More recently, three-dimensionalstable trapping of a microparticle from a single laser beam emitting from asingle-mode fiber with different structure fabricated at the tip of the fiber andwithout any external lens has also been successfully demonstrated [20,21].

Another critical factor that further expands the potential biomedical ap-plications of optical trapping is its capability to measure forces on the order oftens of femto-Newton to hundreds of pico-Newton. This topic will be discussedin greater details in Sect. 14.3 along with potential biomedical applications inSect. 14.4, which together form the core of this chapter.

Other ramifications of optical trapping involve the integration of opti-cal trap with one or more other optical techniques, such as near-field mi-croscopy [22,23], Raman spectroscopy [24,25], and second and third harmonicgeneration microscopy [26], as well as with microfluidics for a wide range ofparticles or cells sorting [27–29].

In this chapter, we focus mainly on the measurement of optical forces in op-tical traps and potential applications of optical trapping as a force transducerfor the measurement of biomolecular forces in the range of sub-pico-Newtonto hundreds of pico-Newton. Partly because of the limited size of this chap-ter, and partly because of our lack of experiences with some ramifications ofthe subjects, several important topics such as those dealing with integratedphoto-voltaic optical tweezers [30,31] and optically driven rotation [32–34] ofmicronsized samples in optical traps are left out. Interested readers are en-couraged to consult excellent accounts of these topics in the references citedabove.

For those who want to learn the subject in greater depth and in moredetails, two excellent review papers [35, 36] each with an impressive listof references can be consulted. In addition, many leading research groupsall over the world have posted incredible amount of information with veryentertaining and informative video demonstrations of a wide range of fea-tures of optical trapping and manipulation. A few selected video demonstra-tions of various forms of optical trapping can also be viewed at our website(http://photoms.ym.edu.tw).

252 A. Chiou et al.

14.2 Theoretical Models for the Calculation of OpticalForces

Optical forces on microparticles in optical traps have been calculated theoreti-cally mainly with the aids of two theoretical models: (1) Ray-Optics Model (orRO Model) [35–38] and (2) Electromagnetic Model (or EM Model) [35,36,39].The former provides a good approximation for cases where the particle sizeis larger than the wavelength of the trapping light, whereas the latter rep-resents a better model for cases where the particle size is smaller than thewavelength of the trapping light (often known as Rayleigh particles). For par-ticle size comparable to the wavelength of the trapping light (often known asMie particles), the calculation is much more involved and the results often lessaccurate. The key concepts of the RO Model and the EM Model are brieflyoutlined below further in this section.

14.2.1 The Ray-Optics (RO) Model

In the RO Model [35–38], each trapping beam is decomposed into a superpo-sition of light rays, each with optical power proportional to the laser beamintensity distribution. For each specific light ray with optical power P in amedium with refractive index n1, propagating along a direction representedby a unit vector ki, the photon momentum per second is given by (Pn1/c)ki,where c is the speed of light in vacuum. When this light ray enter from amedium with refractive index n1 into another medium with refractive in-dex n2, part of the optical power (RP ) is reflected and the remaining part(1 − R)P is refracted (or transmitted) as is prescribed by the Fresnel equa-tions, where R is the optical power reflectance at the interface due to Fresnelreflection. The photon momentum per second associated with the reflectedlight ray is thus (RPn1/c)kr, and that associated with the refracted light rayis [(1 − R)Pn2/c]kt, where kr and kt are the unit vectors representing thepropagation direction of the reflected and the transmitted rays, respectively.The net optical force imparted on the interface due to the photon momentumchange associated with Fresnel reflection (and refraction) of this ray is givenby

F = (Pn1/c)ki − {(RPn1/c)kr + [(1 − R)Pn2/c]kt} (14.1)

or, equivalently,

F = (Pn1/c){ki − [Rkr + (1 − R)(n2/n1)kt]}. (14.2)

In the RO Model, the contribution of the optical force from each constituentray of the trapping beam, as is prescribed by (14.1) or (14.2) above, is addedvectorially to obtain the net optical force on the particle by the beam at eachinterface. For a dielectric microsphere (such as a polystyrene bead or a biolog-ical cell) in water, the difference in refractive indices of the two media is oftenfairly small such that only the momentum change due to the Fresnel reflection


Fig. 14.2. Fresnel reflection and refraction of a light ray at the interface of adielectric microsphere and the surrounding medium in the Ray-Optics (RO) modelof optical trapping

Reflectedlight

Surfaceforce

Incidentlight

Transmittedlight

Laser

Laser Laser

(a)

(b)

Laser Laser

Laser

FnetFnet

Fnet

n1 n2 (>n1)

αα

β

Fig. 14.3. Optical force distribution at the surface of a dielectric microsphere(adapted from Guck et al. [17]) (a) in a single Gaussian beam; (b) in a pair ofcounter-propagating Gaussian beams with equal optical power

at the first (or front) interface and that due to the transmission at the sec-ond (or back) interface are considered to simplify the calculation (Fig. 14.2).In general, optical forces originate from photon momentum changes due toreflection, giving rise to a net force, which mainly pushes the particle alongthe direction of the beam propagation; such a force is often referred to as thescattering force. In contrast, optical forces originating from photon momen-tum changes due to refraction often give rise to a net force pointing towardsthe direction of the gradient of the optical field; such a force is often referredto as the gradient force. Example of theoretical results for the optical forcedistribution on the surface of a dielectric microsphere due to a Gaussian beamand due to a pair of counter-propagating Gaussian beams with equal opticalpower are depicted in Fig. 14.3. Theoretical results from a more recent refined

254 A. Chiou et al.

150

120

90

60

30

0

2

1

z

330

300270

w/R=1

240

210

180

150

120

90

60

30

0

2

1

z

330

300

270

w/R=2

240

210

180

a) b)

Fig. 14.4. Non-uniform force distribution with four distinguished peaks at thesurface of a dielectric microsphere in a pair of counter-propagating Gaussian beamswith equal optical power. Optical power P = 100mW, particle radius R = 3.30 µm,refractive index of surrounding medium n1 = 1.334, refractive index of the particlen2 = 1.378, beam radius = w. (a) w/R = 1; (b) w/R = 2. (adapted from Bareilet al. [38])

model [38], however, indicate several distinguished spikes in the force distri-bution on the surface of a dielectric microsphere in a counter-propagatingdual-beam trap (Fig. 14.4). In the counter-propagating dual-beam trap, theparticle is stably trapped along the axial direction by the balanced scatteringforces from the two beams, and in the transverse direction by the co-operativeaction of the transverse gradient forces from the two beams. In the single-beamgradient-force optical trap, the particle is stably trapped axially, balanced bythe axial scattering force against the axial gradient force, and transversely bythe transverse gradient force.

To compare the efficiency of different optical trapping configurations, theoptical force as given in (14.2) above is often written as

F = (Pn1/c)Q, (14.3)

where the parameter Q, known as the trapping efficiency, represents the frac-tion of photon momentum per second associated with the trapping beam,which is converted into the net trapping force. In most of the optical trappingconfigurations that have been demonstrated to date, the transverse trap-ping efficiency is typically on the order of 0.1–0.001, whereas the axial trappingefficiency is typically a factor of 3–10 smaller than the transverse trappingefficiency. The exact value of the trapping efficiency depends on many fac-tors, including the numerical aperture (NA) of the trapping beam, the size ofthe particle, the refractive index of the particle and that of the surroundingfluid, etc.


Fig. 14.5. (a) The EM model of optical trapping of a dielectric microsphere in aGaussian beam in terms of the minimum potential energy analogous to (b) the forcethat pulls a dielectric block partially filling a parallel plate capacitor connected toa constant voltage source

14.2.2 Electromagnetic (EM) Model

In the EM model [35, 36, 39], stable trapping of a dielectric microparticle inone or more laser beams can be understood in terms of the potential energyminimum of the dielectric particle in the electric field associated with the opti-cal beam. The physical mechanism that a dielectric microparticle is attractedtowards the region with high optical intensity is analogous to the electrostaticforce on a dielectric block partially filling a parallel plate capacitor connectedto a constant voltage source (Fig. 14.5). In cases where the dielectric constantof the particle (or the block in the example of the capacitor) is smaller thanthat of the surrounding medium, the direction of the force is reversed, i.e., theparticle is repelled away from the region of high optical intensity and attractedtowards the region of lower optical intensity. For detailed mathematics of theEM model, please consult the references cited above.

14.3 Experimental Measurements of Optical Forces

Optical forces on microparticles can be measured by several methods. In gen-eral, under identical experimental condition, optical forces scale linearly withoptical power. For a particle in the vicinity of the trap center of a stablethree-dimensional optical trap, the net optical force along the direction ofeach orthogonal axis can be approximated by a Hookean optical spring; a setof optical force constants (or spring constants, kx, ky, and kz) can thus beused as a convenient set of parameters that specifies the three-dimensionaloptical force field within a small volume surrounding each stable equilibriumposition of the particle in the trap. Techniques to measure the optical forcesand optical force constants are described below in this section.

14.3.1 Axial Optical Force as a Function of Position alongthe Optical Axis

One way to measure the axial optical force on a particle in a counter-propagating dual-beam trap as a function of position along the optical axis

256 A. Chiou et al.

is to trap a particle, drive the particle to move back-and-forth along the op-tical axis (by alternately turning on and off one of the trapping beams), andrecord the position of the particle as a function of time z(t) via a precalibratedCCD camera or any other position sensing device. By taking the first and thesecond derivatives of the experimental data z(t) either numerically (or ana-lytically after curve-fitting z(t) with an appropriate polynomial), one obtainsthe velocity v(t) and the acceleration a(t) of the particle. The axial opticalforce Fao(t) can then be determined from the following equation of motion:

ma(t) = Fao(t) − 6πηrv(t) (14.4)

or, equivalently,Fao(t) = ma(t) + 6πηrv(t), (14.5)

where r is the radius of the particle, η is the coefficient of viscosity of the fluidsurrounding the particle, and the mass of the particle m = 4πr3ρ/3, (ρ = thedensity of the particle). The axial optical force as a function of position Fao(z)can be deduced directly from Fao(t) and z(t) by eliminating the parameter tfrom these data.

If we assume a symmetric counter-propagating dual-beam trap with iden-tical optical power from each beam, the axial optical force from each beam willbe identical (except for a sign-change in both the direction of force and thedirection of the relative position of the particle). The net axial optical force,when both beams simultaneously act on the particle, is simply the vector sumof the contribution from each beam. As an example, a set of experimentaldata of z(t) is given in Fig. 14.6a and the axial force deduced by the prescrip-tion described above is given in Fig. 14.6b. In this specific example, the linear

Time (S)

Fo

rce

(pN

)

0.0

0.0−10.0

−60.0 −40.0 −20.0 20.0 40.0 60.00.0

4.0

3.0

2.0

1.0

−1.0

−2.0

−3.0

−4.0

0.0

10.020.030.040.050.060.070.080.090.0

100.0110.0120.0130.0

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Po

siti

on

(m

m)

Position (mm)

(a) (b)

Fig. 14.6. (a) Experimental data showing the axial position of a particle as afunction of time in a counter-propagating dual-beam trap as the particle was drivento move along the optical axis by switching off one of the trapping beams; (b)optical forces on the particle from each beam (represented by the upper and thelower curves) and the net force when both beams were on (represented by the middlecurve) as a function of its relative position along the optical axis. The curves in (b)were deduced from (a) as is described in the text


regime where the axial optical force can be approximated by a linear opticalspring is approximately 80 µm, and the axial optical force constant (given bythe slope of the curve in the linear regime) is approximately 0.05 pN µm−1.Since the total trapping power is approximately 25 mW, the maximum axialtrapping efficiency (at the edge of the linear region) is estimated to be approx-imately 0.02. Although the optically driven motion method described aboveis relatively simple and the experiment can be easily repeated under identicalexperimental condition so that the measurement error can be minimized byaveraging over several sets of data from repeated experiments, it suffers fromseveral shortages: (1) it can not be used for the measurement of transversetrapping force, (2) it assumes that the particle moves linearly along the op-tical axis through out the course of motion, which is a good approximationonly within a certain range ∼100 µm; besides, the linear motion of the parti-cle will be perturbed randomly (and unavoidably) by the thermal fluctuation(Brownian force) and systematically by the gravitation force.

14.3.2 Transverse Trapping Force Measured by Viscous Drag

An earlier method to measure the transverse optical trapping force on a par-ticle in a single-beam gradient-force optical trap (or optical tweezers) is bytrapping a particle and dragging it off transversely with increasing fluid flowspeed until the particle escape from the trap [40]. The fluid flow can be imple-mented either by translating the sample chamber (mounted on a piezoelectric-translational stage) along with the fluid surrounding the trapped particle orwith the aid of a microfluidic pump. At the threshold speed (vth) of thefluid flow when the particle escapes from the optical trap, the transverse op-tical trapping force (Fto) is maximum and is equal to the viscous draggingforce (Fdrag). If the particle radius (r), the coefficient of viscosity (η) of thefluid, and the threshold flow speed (vth) are known, the maximum trans-verse optical trapping force is given by the viscous dragging force; hence,Fto = Fdrag = 6πηrvth (the Stokes’ Law). A systematic error introduced bythe presence of the side wall of the sample chamber (which is ignored in theequation above) can be corrected if the distance of the trapped particle fromthe wall is known [35, 36]. Even though the random error introduced by thethermal fluctuation of the particle can be reduced in principle by statisticallyaveraging over repeated measurements, in practice, the standard deviationsassociated with the experimental results are relatively high (∼50% or higher).A schematic diagram illustrating the underlying principle of this method isgiven in Fig. 14.7, and a video demonstration of this method can be viewedat our website (http://photoms.ym.edu.tw/).

14.3.3 Three-Dimensional Optical Force Field Probed by ParticleBrownian Motion

In the previous sections on the measurement of optical forces, particle fluctu-ation due to Brownian motion was regarded as an unavoidable nuisance that

258 A. Chiou et al.

Fig. 14.7. A schematic illustration of the measurement of transverse optical forceby balancing the optical force against a viscous fluid drag

leads to random errors in the experimental results. Interestingly, the three-dimensional Brownian motion of a particle in an optical trap, when trackedand measured with high precision, can be used to probe the three-dimensionaloptical force field on the particle with relative ease and with fairly high preci-sion. This method, also known as photonics force microscopy [41,42], is brieflydescribed below in this section.

Tracking and the measurement of three-dimensional Brownian motion ofa particle in an optical trap are often accomplished with the aid of one ormore position sensing devices such as a quadrant photo-diode (QPD). Fromthe projection of the particle position fluctuation as a function of time [x(t),y(t), and z(t)] along each orthogonal axis, the corresponding optical forceconstant (kx, ky, and kz) along each axis can be deduced by any one of thethree methods: (1) root-mean-square (rms) fluctuation of the particle position[36], (2) the Boltzmann distribution of the particle in a parabolic potentialwell [41,42], (3) the temporal frequency analysis of particle position fluctuation[43,44]. To limit the length of this chapter, only the second approach based onBoltzmann distribution of the particle in a parabolic potential well is discussedbelow in this section.

An experimental setup used by the authors to probe the three-dimensionaloptical force field on a particle trapped in a fiber-optical counter-propagatingdual-beam trap is shown in Fig. 14.8 [14]. A laser beam (cw, λ = 532 nm froma Nd:YVO4 laser) was expanded and collimated via a 3X beam-expander (3XBE) through a half-wave plate (λ/2) and a polarizing beam splitter (PBS)cube to split into two beams with equal optical power, and each coupled intoa single-mode fiber (NA = 0.11) via a single-mode fiber coupler (SMFC). Theoutput ends of the two fibers were aligned so that the two laser beams exitingfrom the fibers formed a pair of counter-propagating beams along a commonoptical axis inside a sample chamber where microparticles or living cells weretrapped in PBS solution. The distance between the two fiber end-faces waskept at 125 µm. Portions of the trapping beams scattered by the trapped par-ticle were collected by a pair of orthogonally oriented long-working distantobjectives (LOBI and LOBII, ×50, NA = 0.42) and projected onto a pair ofquadrant photodiodes (QPDI and QPDII). LOBI/QPDI and LOBII /QPDIIcollected the scattered lights and tracked the position of the trapped particle


Nd : YVO4 λ=532nm

3X BE PBSλ /2SMFC

SMFCLOB

LOB

QPD

QPD

CCD

BS

Lamp

z

y

x

Fig. 14.8. A schematic illustration of the experimental setup for the mapping ofthree-dimensional optical force field on a particle in a fiber-optical dual-beam trap

projected on the xz-plane and the yz-plane, respectively (the xyz co-ordinatesystem is depicted in the upper left corner in Fig. 14.8). Besides, the LOBI wasalso used for imaging the trapped particle on a CCD camera (752×582 pixels)via incoherent illumination from a lamp. To obtain the conversion factor thatconverts the output voltage of the QPD into the particle displacement, wetrapped the microparticle of interest on the optical axis in the middle of theend-faces of two optical fibers via equal laser power from the two fibers, andmomentarily turned off one of the laser beams to drive the particle (by theremaining single beam) along the optical axis towards the opposite end-faceof the optical fiber. The second beam was subsequently turned on to drivethe particle back along the optical axis towards the original equilibrium po-sition at the middle of the fiber end-faces. As the microparticle was drivenback along the optical axis under the illumination of both beams, the corre-sponding output voltage of the quadrant photodiode and the position of themicroparticle on the CCD camera were recorded simultaneously. An illustra-tive example of such a calibration curve (i.e., QPD output voltage vs. particleposition recorded on the CCD) is depicted in Fig. 14.9, which shows a lineardependence with a slope of approximately 0.986 µm V−1 within the regionof approximately 3.4 µm. Besides the Brownian fluctuation, the accuracy ofthis approach is mainly limited by the determination of the particle positionfrom the incoherent image of the illuminated particle on the pixelated dig-ital CCD camera. The calibration method described above was carried outduring the return trip of the particle (from a point offset from the centerto the center equilibrium position) when both laser beams were on to ensurethat the particle was evenly illuminated from both sides. Recently the authorshave tracked the three-dimensional Brownian motion of polystyrene and silica

260 A. Chiou et al.

Fig. 14.9. An example of a calibration curve for the conversion of the QPD outputvoltage into the axial position of a particle in a fiber-optical dual-beam trap

beads of different sizes as well as Chinese hamster ovary cells trapped in afiber-optical dual-beam trap, and analyzed their position distribution to ob-tain the optical force field approximated by a three-dimensional parabolicpotential well [14, 15]. Under the parabolic potential approximation and theclassical Boltzmann statistics, the associated optical force constant along eachaxis can be calculated from the following two equations [41]:

ρ (z) = C exp [−E (z)/KBT ] , (14.6)

E (z) = −KBT ln ρ (z) + KBT lnC = kzZ2/2, (14.7)

where ρ(z) is the probability function of the trapped particle position along thez-axis, C is the normalization constant, E(z) is the potential energy functionalong the z-axis, KB is the Boltzmann constant, and kz is the optical forceconstant along the z-axis. Identical set of equations also apply for the x-axisand the y-axis. As a specific example, a set of experimental data representingthe parabolic potential E(x) and E(z) of the optical force field on a 2.58 µmdiameter silica particle in a fiber-optical dual-beam trap (total trapping power= 22 mW, distance between the fiber end-faces = 125 µm,) is depicted inFig. 14.10. The solid lines represent the theoretical curves based on (14.7)given above; the corresponding optical force constants, defined by (14.7), werekx = 0.16 pN µm−1 and kz = 0.04 pN µm−1. The force constant along theoptical axis is weaker than those along the transverse axes, which is consistentwith the theoretical results reported earlier [45]. This is also true in the case ofoptical tweezers [46]. For clarity sake, the experimental data for y-axis along


Distance from the trapping center (m)-1.3E-7

-3.5E-20

-3.0E-20

-2.5E-20

-2.0E-20

-1.5E-20

-1.0E-20

-5.0E-21

0.0E+0

1.3E-7-5.0E-8 5.0E-80.0E+0

Op

tica

l Po

ten

tial

En

erg

y (J

)

Fig. 14.10. Experimental data representing the parabolic optical potentials E(x)(the inner set of data points with a steeper slope) and E(z) (the outer set of datapoints) on a 2.58 µm silica particle. The solid lines represent the theoretical fits basedon (14.7)

with the theoretical fit with ky = 0.15 pN µm−1 are not shown in Fig. 14.10;the data and the theoretical curve for y-axis essentially overlap with those forthe x-axis.

14.3.4 Optical Forced Oscillation

Optical forced oscillation of a trapped particle is another powerful techniquefor the measurement of optical force and biological forces such as those as-sociated with protein–protein interaction. Optical forced oscillation (OFO)refers to the trapping and forced oscillation of a microparticle in an opticaltrap [47, 48]. Transverse force constants of the optical trap can be measuredwith fairly high precision by measuring the oscillation amplitude and the rel-ative phase (with respective to that of the driving source) of the particle asa function of oscillation frequency. OFO can be used as a convenient tool forthe measurement of protein–protein interaction [49, 50] and of protein–DNAinteraction with the aid of a microparticle coated with appropriate protein ofinterest. In this section, we introduce the basic principle and the experimentalimplementation of optical forced oscillation.

Optical forced oscillation has been implemented by one of the two follow-ing methods: (1) by trapping a microparticle in a conventional single-beamgradient-force optical trap (or optical tweezers) and scanning the trappingbeam sinusoidally with a constant amplitude and frequency (Fig. 14.11a),(2) by trapping a microparticle in a set of twin optical tweezers and chop-ping one of the trapping beams on-and-off at a constant chopping frequency

262 A. Chiou et al.

Fig. 14.11. A schematic diagram of optical forced oscillation via (a) oscillatoryoptical tweezers and (b) a set of twin optical tweezers

(Fig. 14.11b). In both cases, the steady-state oscillation amplitude and therelative phase (with respect to that of the driving source) of the oscillatingparticle can be conveniently measured with the aid of a quadrant photo-diodein conjunction with a lock-in amplifier. By changing the driving frequency,typically in the range of approximately a few hertz to a few hundred hertz,the oscillation amplitude and the relative phase of the oscillating particle canbe plotted as a function of frequency. In general, the experimental data fitfairly nicely with the theoretical results deduced from a simple theoreticalmodel of forced-oscillation with damping [48, 50]. By coating a microparticlewith a protein of interest and allowing the oscillating particle to interact withprotein receptors on a cellular membrane, the method described above can beused to measure the protein–protein interaction, which can be modeled as an-other linear spring in parallel to the optical spring (Fig. 14.12). The equationof motion of a particle, suspended in a viscous fluid, trapped, and forced tooscillate in oscillatory optical tweezers, can be written as

mx = −βx − k[x (t) − A

(eiωt

)], (14.8)

where m is the mass of the particle, βx = 6πηrx is the viscous drag prescribedby Stokes’ law, η is the coefficient of viscosity of the surrounding fluid, r isthe radius of the particle, k is the optical spring constant in the linear springmodel, and A and ω are the amplitude and the frequency of the focal spot ofthe oscillatory optical tweezers, respectively.

In the case of a set of twin optical tweezers with the particle interactingwith a cell as is illustrated schematically in Fig. 14.12, the equation of motionof the particle can be written as [48]

mx = −βx − k(eiωt + 1)2

(x − x1) − (k + kint)(x − x2), (14.9)

where x1 and x2 represent the position of the force center of optical tweezers 1and 2, respectively; ω is the fundamental harmonics of the chopping frequency;


YY

B2

k2

kint

x2x1

k2

kint

x2x1

k2

kint

x2x1

k2

kint

Y

B1

k1

k2

kintm

x2x1

Fig. 14.12. A simplified linear spring model of a particle simultaneously acted uponby optical forces from a set of twin optical tweezers and a cellular interactive forcewhen a cell is in contact with the particle in the equilibrium trapping position ofthe tweezers on the right

kint is force constant of the protein–protein interaction modeled as a linearspring; and the rest of the symbols are the same as those described earlierin association with (14.8). In (14.9), we have assumed that the equilibriumposition of the protein–protein interaction at the cellular membrane coincideswith that of the second optical tweezers; this was accomplished experimentallyby bringing the cell to touch the particle when it was stably trapped in thesecond tweezers as is illustrated schematically in Fig. 14.12.

The steady-state solution at the fundamental frequency of (14.8) and(14.9) above can be expressed as

x (t) = D[ei(ωt−φ)

](14.10)

andx (t) = D

[ei(ωt−φ)

]− x0, (14.11)

respectively, where D is the amplitude, φ the phase lag (with respective tothat of the driving source), and x0 the equilibrium position of the oscillatingparticle.

Solving (14.8) with an assumed solution in the form of (14.10) gives

D (ω)D (0)

=k√

k2 + (βω)2 − mω2

(14.12)

264 A. Chiou et al.

and

φ (ω) = tan−1

(βω

k − mω2

). (14.13)

Likewise, solving (14.9) with an assumed solution in the form of (14.11) gives

D(ω)D(0)

=32k′√

(k′ − mω2)2 + (βω)2(14.14)

andφ(ω) = tan−1

{βω

/[k′ − mω2

]}, (14.15)

where k′ = (3/2)k + kint.Experimental data representing the amplitude and the relative phase of

the oscillation of a trapped 1.5 µm polystyrene particle in water as a functionof frequency along with the corresponding theoretical fits based on (14.12)and (14.13) given above are depicted in Fig. 14.13a,b, respectively. The op-tical spring constant kx was deduced from the best fit to be 7.69 pN µm−1,from the amplitude data, and 7.61 pN µm−1, from the phase data. The opti-cal spring constants measured by different methods described above agree towithin approximately 8%. In contrast to the Brownian motion method, the re-sults obtained by optical forced oscillation method depend only on the relativeamplitude of particle oscillation and not on its absolute value. Besides, thesignal-to-noise ratio is significantly enhanced by the phase locking technique.Although the transverse optical spring constant along any specific directioncan be measured via oscillatory optical tweezers with reasonable accuracy byscanning the trapping beam along the selected direction, the two-dimensional

Oscillation Frequency (Hz) Oscillation Frequency (Hz)

Rel

ativ

e P

has

se (

deg

ree)

1.00.5 −5.0

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

50.0

0.6

0.7

0.8

0.9

1.0

1.1

10.0 100.0 1.0 10.0 100.0

(a) (b)

No

rmal

ized

Am

plit

ud

e

Fig. 14.13. (a) The normalized amplitude, (b) the relative phase of the opti-cal forced oscillation of a polystyrene particle (diameter = 1.5 µm) suspended indeionized water as a function of the oscillation frequency. The solid curves are the-oretical fits


Table 14.1. A comparison of optical spring constants measured by differentmethods

Method Analysis kOT(x) kOT(y)(pN µm−1) (pN µm−1)

Brownian motion Displacement variance 7.07 6.31Potential well 7.28 6.84Power spectrum 7.21 6.88

Optical forced oscillation Amplitude 7.69Phase 7.61

optical force field E(x, y) can be simultaneously mapped, and the associatedspring constants kx and ky conveniently measured in stationary optical tweez-ers via the Brownian motion method. The optical spring constants kx and ky

on the transverse plane obtained from the Brownian motion analysis agreewith each other to within approximately ±6%, which is consistent with theearlier theoretical and experimental results [46, 51, 52] and also with thosereported previously for the case of a fiber-optical dual-beam trap [14,45].

As an example, transverse optical spring constants, kx and ky, for apolystyrene particle (diameter = 1.5 µm) suspended in deionized water andtrapped in optical tweezers (characterized by λ = 1064 nm, NA = 1.0, trap-ping optical power = 2 mW) obtained by different methods are compared andsummarized in Table 14.1.

14.4 Potential Biomedical Applications

Potential biomedical applications of optical trapping and manipulation in-clude (1) trapping and stretching a cell to measure its visco-elastic propertyand to correlate with its physiological condition, (2) trapping two cells to mea-sure cell–cell interaction, (3) trapping two protein-coated microparticles tomeasure protein–protein interaction, (4) trapping one protein-coated particleto interact with membrane proteins on a cell to measure the protein–proteininteraction at the cellular membrane, (5) trapping two beads with a segmentof DNA molecule stretched in between to measure the interaction dynamicsof the DNA with proteins injected into the sample chamber. Selected videodemonstrations of some of the features listed above can be viewed at thewebsite of the authors’ lab (http://photoms.ym.edu.tw).

In general, optical trapping and manipulation promise to provide one ormore of the following unique features in biomedical applications:

1. Micropositioning : One can guide cell–cell, cell–molecule, or molecule–molecule interactions in terms of where and when the interactionstake place. The interaction can therefore be measured within a timepoint on the order of “second” after the initiation of the molecularinteraction.

266 A. Chiou et al.

2. Living cell capability : The setup is readily adaptable for living cellimaging and manipulation.

3. The technique can be easily integrated with laser spectroscopy andmicroscopy for single cell measurement with subcellular resolution.

4. Single molecule resolution: Several protocols have been developed tomeasure the force exerted by a single biomolecule or the interactionbetween a molecular pair. New methods and algorithms have beendeveloped enabling the extraction of single molecular results from thebulk measurements or other population data.

5. The assay is sensitive to the functions of the biomolecules: One canmonitor in real-time the dynamics of molecular binding or of biologicalforces associated with molecular interactions to examine the functionsof the biomolecules.

6. The assay is sensitive to the conformation of the biomolecules: Onemay obtain structural information of the biomolecule by integrat-ing optical tweezers with other photonics modalities (such as time-resolved fluorescence microscopy and Raman spectro-microscopy).

7. The possibility of noncontact and noninvasive cell compliance mea-surements may lead to a new paradigm for assessing the physiologyand the pathology of living cells with potential clinical applications.

In this section, we outline a few selected examples of the applications ofoptical trapping for protein–protein interactions, protein–DNA interactions,and cellular trapping and stretching.

14.4.1 Optical Forced Oscillation for the Measurementof Protein–Protein Interactions

As an example of protein–protein interaction, we present in this section theapplication of a set of twin optical tweezers to trap and oscillate a ConA(lectin)-coated polystyrene bead and to measure its interaction with glycopro-tein receptors at the cellular plasma membrane of a Chinese hamster ovary(CHO) cell [50]. The bead was trapped between two quadratic potential wellsdefined by a set of twin optical tweezers and was forced to oscillate by chop-ping on-and-off one of the trapping beams. We tracked the oscillatory motionof the bead via a quadrant photodiode and measured with a lock-in amplifierthe amplitude of the oscillation as a function of frequency at the fundamen-tal component of the chopping frequency over a frequency range from 10to 600 Hz. By analyzing the amplitude as a function of frequency for a freebead suspended in buffer solution without the presence of the CHO cell andcompared with the corresponding data when the bead was interacting withthe CHO cell, we deduced the transverse force constant associated with theoptical trap and that associated with the interaction by treating both theoptical trap and the interaction as linear springs. The force constants weredetermined to be approximately 2.15 pN µm−1 for the trap and 2.53 pN µm−1


Time (s)

0

1

2

3

4

5

6Con A coated beadBSA coated bead

Rel

ativ

e O

scill

atio

n A

mp

litu

de

(V)

0 5 10 15 20 25 30 35 40 45 50 55

Fig. 14.14. The time-dependence of the relative amplitude of a polystyrene bead(diameter = 2.83 µm) executing optical forced oscillation at 50Hz in the vicinityof a CHO cell. The experimental data for ConA-coated bead are denoted by solidsquares and the data for a BSA-coated bead are denoted by “*”

for the lectin–glycoprotein interaction. When the CHO cell was treated withlantrunculin A, a drug that is known to destroy the cytoskeleton of the cell,the oscillation amplitude increased with time, indicating the softening of thecellular membrane, until a steady state with a smaller force constant wasreached. The steady state value of the force constant depended on the drugconcentration.

As an illustrative example, the time dependence of the relative amplitudeof a polystyrene bead (diameter = 2.83 µm) oscillating at 50 Hz in the vicinityof a Chinese hamster ovary cell is depicted in Fig. 14.14 for the case of a BSA-coated bead and that of a ConA-coated bead. The decay of the oscillationamplitude in the case of the ConA-coated bead is a manifestation of theinteraction of ConA protein with the glycol-protein on the cellular membraneof the CHO cell.

14.4.2 Protein–DNA Interaction

The interaction of proteins with DNA plays a pivotal role on a number ofimportant biological processes, including DNA repair, replication, recombina-tion, and segregation. Many different proteins can be used as models for suchinvestigations. For example, one can optically trap-and-stretch a segment ofdsDNA to analyze the dynamics of homologous DNA search and strand ex-change reaction mediated by RecA protein.

In the study of protein–DNA interaction, micron-size polystyrene beadsare often attached to the ends of each DNA sample, one at each end, to serveas handles for optical tweezers to trap and stretch the DNA sample betweenthe beads. A schematic illustration of a dsDNA segment stretched betweena fixed (large) bead (with diameter ∼20 µm) and an optically trapped small

268 A. Chiou et al.

Opticaltweezers

polystyrene particle ~20µµµµm

Unwinding dsDNA

RecA binding ssDNAd(t)

polystyreneparticle ~20µm

polystyrene particle~2µm

Fig. 14.15. A schematic illustration of a segment of a DNA molecule stretchedbetween two polystyrene beads (with one bead fixed to the bottom of the samplechamber and the other bead trapped in optical tweezers) for the measurement ofthe dynamics of the interaction of the DNA sample with RecA proteins injected intothe sample chamber

bead (with diameter ∼1 µm) for the study of the dynamic of the interactionwith RecA-proteins carrying complementary segment of ssDNA is shown inFig. 14.15. During the interaction, the stretched length and the elastic con-stant of the DNA can be measured in real-time by monitoring either the po-sition of the trapped particle as a function of time (in the case of stationaryoptical tweezers) or the amplitude and the phase of the oscillating particleas function of frequency and time (in the case of oscillatory optical tweezers)(Fig. 14.16). One of the technical challenges of this approach is the lack of aneffective method to avoid the adherence of multiple-strings of DNA moleculesin parallel between the two beads. Although this can be achieved in principleby optimizing the ratio of the concentration of the DNA samples and that ofthe beads during the sample preparation, it is very tedious and inefficient inpractice.

In the steady state, the bead is expected to wonder around an equilibriumposition (slightly displaced from the optical axis) dictated by the balancingof the transverse gradient optical force and the elastic force of the stretchedDNA molecule. The position fluctuation of the bead is a manifestation of theBrownian force acting on the bead and on the DNA molecule as well as theconformational change of the DNA molecule. Specifically, the stretching or thecontraction as well as the winding or the unwinding of the double-strand DNAmolecule is revealed by the translational and the rotational motion of the beadof which the former can be measured via a quadrant photodetector (QPD) orany other optical position sensing detector, while the later can also be mea-sured optically by using a bead with optical birefringence in conjunction with


Fig. 14.16. A schematic illustration of the measurement of the dynamics of DNA–protein interaction by stretching the DNA between two beads, trapping one of thebead, and tracking its Brownian motion with a quadrant-photodiode

any polarization sensitive detection scheme. Instead of a bead with opticalbirefringence, any bead lacking axial symmetry illuminated by a laser beam isalso expected to generate an optical scattering signal modulated (in intensity)with a frequency component at that of the rotational frequency of the bead.The basic principle of this approach is to detect (by analyzing the transla-tional and the rotational motion of the trapped bead) the dynamic of theconformational change of a stretched double-strand DNA molecule interact-ing with RecA-ssDNA filaments injected into the surrounding buffer solution.The generic goal of the experiments, such as the one outlined above, is to un-derstand the relationship between the physical properties and the biochemical(or functional) properties of DNA at a fundamental level.

14.4.3 Optical Trapping and Stretching of Red Blood Cells

As mentioned earlier, a red blood cell (RBC) can be trapped and stretchedwith the aid of a fiber optical dual-beam trap-and-stretch to measure its visco-elastic property. A schematic diagram to illustrate the application concept ofa fiber-optical dual-beam trap-and-stretch in conjunction with a microfluidicflow chamber, fabricated with poly dimthylsiloxane (PDMS), to inject theRBC one at a time for such measurement is shown in Fig. 14.17. Photographsof an experimental set up in our laboratory are depicted in Fig. 14.18. Pre-liminary experimental results illustrating the morphological change of humanRBC samples, osmotically swollen into spherical shape, as a function of theoptical power are shown in Fig. 14.19 along with the experimental data onthe fractional change in length of the major axis and the minor axis of a

270 A. Chiou et al.

LaserLaserLaser

Filter

Optical Coupler splitter

60X NA=0.85

Objective

CCD

LaserLamp

= 1064nm

Single Model fiberFlow Chamber

Top view of the flow chamber; material: Poly Dimthylsiloxane (PDMS)

Fig. 14.17. A schematic illustration of the experimental setup for simultaneoustrapping, stretching, and morphological deformation measurement of red blood cellsin a fiber-optical dual-beam trap

Fig. 14.18. Photographs of the experimental setup illustrated schematically inFig. 14.17

human RBC sample as a function optical power. From these experimentaldata, the product (Eh) of the elasticity (E) and the thickness (h) of the RBCcell membrane was estimated to be Eh = 2.4× 10−4 N m−1, with the aid of asimple theoretical model [15–17].


19mw 43mw 66mw 92mw 118mw 146mw 174mw

Eh=2.4¥10−4 N/m

RBC Diameter : 6.5±0.5mm

peak stress (N/m2)

y=0.0164x−0.0069R

2=0.9808

y=−0.0157x−0.0077R

2=0.9782

0 0.1

−0.1

−0.15

−0.05

0.05

0.15

0.1

0

0.2 0.3 0.4 0.5 0.6 0.7

∇ ρ/ρ

∇

ρ (0)/ρ (0)∇

ρ(π/2)/ρ (π/2)

fit linefit line

Fig. 14.19. Photographs of a human red blood cell (RBC), osmotically swolleninto spherical shape, trapped and stretched at different laser power in a fiber-opticaldual-beam trap, along with the experimental data showing the relative change inlength along the major and the minor axes of the RBC sample as a function of laserpower

14.5 Summary and Conclusion

In this chapter, we give a brief historical overview of optical trapping andmanipulation followed by an outline of two theoretical models, namely theray optics model and the electromagnetic model, along with several methodsfor the experimental measurement of optical forces with sub-pico-Newton res-olution. Unique features of optical trapping and manipulation for biomedicalapplications are outlined and examples including the measurement of the dy-namics of protein–protein interaction and protein–DNA interaction as well asthe simultaneous trapping, stretching, and morphological deformation mea-surement of human red blood cells are highlighted.

Acknowledgment

Research in optical trapping and manipulation carried out by the authors issupported by the National Science Council of the Republic of China GrantsNSC 95-2752-E010-001-PAE, NSC 94-2120-M-010-002, NSC 94-2627-B-010-004, NSC 94-2120-M-007-006, NSC 94-2120-M-010-002, and NSC 93-2314-B-010-003, and the Grant 95A-C-D01-PPG-01 from the Aim for the TopUniversity Plan supported by the Ministry of Education of the Republic ofChina.

References

1. A. Ashkin, Phys. Rev. Lett. 24, 156 (1970)2. A. Ashkin, J.M. Dziedzic, Appl. Phys. Lett. 19, 283 (1971)3. A. Ashkin, J. Dziedzic, J. Bjorkholm, S. Chu, Opt. Lett. 11, 288 (1986)

272 A. Chiou et al.

4. A. Ashkin, J.M. Dziedzic, Nature 330, 769 (1987)5. M.M. Burns, J.M. Fournier, J.A. Golovchenko, Science 249, 749 (1990)6. A. Chiou, W. Wang, G.J. Sonek, J. Hong, M.W. Berns, Opt. Commun. 133, 7

(1997)7. K. Sasaki et al., Jpn. J. Appl. Phys. 30, L907 (1991)8. E. R. Dufresne, D.G. Griera, Rev. Sci. Instrum. 69, 1974 (1998)9. E.R. Dufresne, G.C. Spalding et al., Rev. Sci. Instrum. 72, 1810 (2001)

10. G. Sinclair, P. Jordan et al., Opt. Exp. 12, 5475 (2004)11. R.L. Eriksen, V.R. Daria, J. Gluckstad, Opt. Exp. 10, 597 (2002)12. P.J. Rodrigo, I.R. Perch-Nielsen et al., Opt. Exp. 14, 13107 (2006)13. A. Constable, J. Kim, J. Mervis, F. Zarinetchi, M. Prentiss, Opt. Lett. 18, 1867

(1993)14. M.T. Wei, K.T. Yang, A. Karmenyan, A. Chiou, Opt. Exp. 14, 3056 (2006)15. J. Guck, R. Ananthakrishnan, T.J. Moon, C.C. Cunningham, J. Kas, Phys. Rev.

Lett. 84, 5451 (2000)16. J. Guck, R. Ananthakrishnan, H. Mahmood, T.J. Moon, C.C. Cunningham,

J. Kas, Biophys. J. 81, 767 (2001)17. J. Guck, R. Ananthakrishnan, C.C. Cunningham, J. Kas, J. Phys. Cond. Mat.

14, 4843 (2002)18. J. Guck, S. Schinkinger, B. Lincoln et al., Biophys. J. 88, 3689 (2005)19. W. Wang, A. Chiou, G.J. Sonek, M.W. Berns, J. Opt. Soc. Am. B 14, 697

(1997)20. R.S. Taylor, C. Hnatovsky, Opt. Exp. 11, 2775 (2003)21. S.I. Eom, Y. Takaya, T. Miyoshi, T. Hayashi, Proc. SPIE 6326, 70 (2006)22. M. Gu, P. Ke, Opt. Lett. 24, 74 (1999)23. T. Sugiura, S. Kawata et al., J. Microsc. 194, 291 (1999)24. K. Ramser, K. Logg et al., J. Biomed. Opt. 9, 593 (2004)25. A. Fontes, K. Ajito et al., Phys. Rev. E 72, 012903 (2005)26. E.V. Perevedentseva, A.Y. Karmenyan, F.J. Kao, A. Chiou, Scanning 26(Suppl.

I), 178 (2004)27. R.W. Applegate Jr., J. Squier, Opt. Exp. 12, 4390 (2004)28. S.C. Chapin, V. Germain et al., Opt. Exp. 14, 13095 (2006)29. P.T. Korda, M.B. Taylor et al., Phys. Rev. Lett. 89, 128301 (2002)30. P.Y. Chiou, A.T. Ohta, M.C. Wu, Nature 436, 370 (2005)31. A.T. Ohta, P.Y. Chiou, M.C. Wu, Proc. SPIE 6326, 632617 (2006)32. V. Bingelyte, J. Leach et al., Appl. Phys. Lett. 82, 829 (2003)33. A.I. Bishop, T.A. Nieminen et al., Phys. Rev. Lett. 92, 198104 (2004)34. M. Gudipati, J.S. D’Souza et al., Opt. Exp. 13, 1555 (2005)35. A. Ashkin, IEEE J. Sel. Top. Quantum Electron. 6, 841 (2000)36. K.C. Neuman, S.M. Block, Rev. Sci. Instrum. 78, 2787 (2004)37. A. Ashkin, Biophys. J. 61, 569 (1992)38. P.B. Bareil, Y. Sheng, A. Chiou, Opt. Exp. 14, 12503 (2007)39. Y. Harada, T. Asakura, Opt. Commun. 124, 529 (1996)40. W.H. Wright, G.J. Sonek, M.W. Berms, Appl. Phys. Lett. 63, 715 (1993)41. E.L. Florin, A. Pralle, E.H.K. Stelzer, J.K.H. Horber, Appl. Phys. A 66, 75

(1998)42. A. Rohrbach, H. Kress, E.H. Stelzer, Opt. Lett. 28, 411 (2003)43. L.P. Ghislain, W.W. Webb, Opt. Lett. 18, 1678 (1993)44. L.P. Ghislain, N.A. Switz, W.W. Webb, Rev. Sci. Instrum. 65, 2762 (1994)


45. E. Sidick, S.D. Collins, A. Knoesen, Appl. Opt. 36, 6423 (1997)46. M.T. Wei, A. Chiou, Opt. Exp. 13, 5798 (2005)47. L.A. Hough, H.D. Ou-Yang, J. Nanopart. Res. 1, 495 (1999)48. M.T. Wei et al., Proc. SPIE 6447, 06 (2007)49. M.T. Wei, K.F. Hua, J. Hsu, A. Karmenyan, H.Y. Hsu, A. Chiou, Proc. SPIE

6326, 63260 (2006)50. S.L. Liu et al., Opt. Exp. 15, 2713 (2007)51. C. Xie, M.A. Dinno, Y.Q. Li, Opt. Exp. 13, 1621 (2005)52. A. Rohrbach, Phys. Rev. Lett. 95, 168102 (2005)

15

Laser Tissue Welding in Minimally InvasiveSurgery and Microsurgery

R. Pini, F. Rossi, P. Matteini, and F. Ratto

15.1 Introduction

Lasers have become already a widespread tool in many operative and ther-apeutic applications in the surgical field. The so-called “minimally invasive”laser techniques provide remarkable improvements: in these, laser surgery isperformed inside the human body through small incisions by means of opticalfibre probes and endoscopes, or laser tools are proposed as a replacement forconventional tools to minimize the surgical trauma, such as in the case oflaser-induced suturing of biological tissues. The aim of these procedures is toimprove the quality of life of patients, by decreasing healing times and therisk of postoperative complications.

Joining tissue by applying laser irradiation was first reported at the endof the 1970s, when a neodymium/yttrium–aluminium–garnet (Nd:YAG) laserwas used for the microvascular anastomosis of rat carotid and femoral arteries.Ever since, laser tissue welding has been evaluated in several experimentalmodels including blood vessels, skin, nerve, intestine, uterine tube, and soon [1, 2]. Laser welding has progressively assumed increased relevance in theclinical setting, where it appears to be a valid alternative to standard surgicaltechniques. At present, there are many applications of tissue welding that arebeginning to achieve widespread acceptance.

Many types of lasers have been proposed for laser tissue welding. Infraredand near-infrared sources include carbon dioxide (CO2), thulium–holmium–chromium, holmium, thulium, and neodymium rare-earth-doped-garnets(THC:YAG, Ho:YAG, Tm:YAG, and Nd:YAG, respectively), and galliumaluminium arsenide diode (GaAlAs) lasers. Visible sources include potassium-titanyl phosphate (KTP) frequency-doubled Nd:YAG, and argon lasers.The laser energy is absorbed by water at the infrared wavelengths and byhaemoglobin and melanin at the visible wavelengths, thereby producing heatwithin the target tissue. As the temperature rises, the extracellular matrix ofthe connective tissue undergoes thermal changes that lead to the welding ofthe wound (Fig. 15.1).

276 R. Pini et al.

Fig. 15.1. Structural modifications induced in fibrillar collagen of connective tissueby temperature rise. Normal triple helix collagen molecules are packed in a quarter-staggered manner and connected by covalent bonds to form a collagen fibril (left).When heat is applied, hydrolysis of intramolecular hydrogen bonds occurs, whichresults in the unwinding of the triple helices (middle). The first step (1) leads to ashrinkage effect parallel to the axis of the fibrils. At higher temperatures, covalentcross-links connecting collagen strands break, resulting in a complete destruction ofthe fibrillar structure (right) and causing relaxation of the tissue (step 2)

Laser tissue welding has been shown to possess several advantages com-pared with conventional closure methods, such as reduced operation times,fewer skill requirements, decreased foreign-body reaction and thereforereduced inflammatory response, faster healing, increased ability to induceregeneration, and an improved cosmetic appearance. Laser welding also hasthe potential to form complete closures, thus making possible an immediatewatertight anastomosis, which is particularly important in the case of vas-cular, genito-urinary tract, and gastrointestinal repairs. A watertight closurealso discourages the exit of regenerating axons and the entry of fibroblasts.Lastly, laser welding can be used endoscopically and laparoscopically to ex-tend the range of its applications to cases in which sutures or staples cannotbe used.

However, despite the large number of experimental studies reported in theliterature, very few of them have reached the clinical phase. This is mainly be-cause of the lack of clear evidence of the advantages of laser-assisted suturingagainst conventional methods, and because of a low reproducibility of results.The damage induced in tissues by direct laser heating and heat diffusion andthe poor strength of the resulting welding are the main problems as far as

15 Laser Tissue Welding in Minimally Invasive Surgery and Microsurgery 277

future clinical applications of the laser-assisted procedure are concerned. Infact, as water, haemoglobin, and melanin are the main absorbers of laser lightwithin tissue, the heating effect is not selectively limited to a target area, andall irradiated tissues are heated. For instance, the CO2 laser has been used forlaser repairs of thin tissues because of its short penetration depth (<20 µm).However, for thicker tissues, welding has been achieved only by irradiatingwith high laser power and longer exposure times, thus inducing high levels ofheat damage [3]. The emissions of other near-infrared lasers, such as Nd:YAGand diode lasers, are more suited to the welding of thicker tissues. In any case,control of the dosimetry of laser irradiation and of corresponding temperaturerise is crucial to minimize the risk of heat damage to the tissue and to generatestrong welds.

Two advances have been useful in addressing the issues associated withlaser tissue welding: the application of laser-wavelength-specific chromophoresand the addition of endogenous and exogenous material to be used as solder.

The use of wavelength-specific chromophores enables differential absorp-tion between the stained region and the surrounding tissue. The advantageis primarily a selective absorption of laser radiation by the target, withoutthe need for a precise focusing of the laser beam. Moreover, lower laser ir-radiances can be used because of the increased absorption of stained tissues.Various chromophores have been employed as laser absorbers, including in-docyanine green (ICG) [4], fluorescein [5], basic fuchsin, and fen 6 [6]. Theuse of a near-infrared laser – which is poorly absorbed by biological tissues –in conjunction with the topical application of a dye with an absorption peakoverlapping the laser emission, is a very popular setting for the laser weldingtechnique. Diode lasers emitting around 800 nm and ICG have been used incorneal tissue welding in cataract surgery and corneal transplant [7,8], vascu-lar tissue welding [9–11], skin welding [4, 12], and in laryngotracheal mucosatransplant [13].

Laser welding by means of solders, namely “laser soldering,” makes useof exogenous solders as topical protein preparations. This makes possible abonding of the adjoining and underlying tissues when activated by laser light.The extrinsic agents provide a large surface area over which fusion with thetissue can occur, thus favoring the approximation of the wound edges thateventually heal together in the postoperative period. Useful welding materialsinclude blood [14], plasma [15], fibrinogen [16] and albumin, which is theone most frequently employed [5,17]. Several studies have demonstrated thatthe addition of an albumin solder to reinforce laser tissue repairs significantlyimproves postoperative results [5,18]. Moreover, incorporation into the proteinsolder of a laser-absorbing chromophore makes it possible to confine the heatinto the area of solder application, which reduces the extent of collateral heatdamage to adjacent tissues. ICG-doped albumin has become an increasinglypopular choice in the last decade [17]. The laser is used to denature the proteinimmediately after application of the protein solder to the wound site, thusyielding a bond at the solder–tissue interface.

278 R. Pini et al.

Another improvement of traditional protein solders relies on the use of syn-thetic polymers mixed with serum albumin. This provides better flexibility,as well as improved repair strength over albumin-protein solders alone [19].Poly(lactic-co-glycolic acid) (PLGA) is a class of synthetic biodegradablepolymers that are easily degraded in vivo and are eliminated through normalmetabolic pathways. These materials can be employed also as drug-deliverysystems by adding a range of dopants including antibiotics, anaesthetics, andvarious growing factors that may enhance the rate and quality of wound heal-ing [20].

A more recent development in laser tissue welding is the use of an infraredtemperature feedback control of the laser device. Diagnostic feedback via sur-face temperature measurements has been used to control laser power deliveryto achieve well-defined welding protocols [21]. Although surface temperaturemonitoring may be important for a primary control of the energy delivery,it does not reveal dynamic changes occurring below the tissue surface, whichis where the weld actually occurs. This problem has been addressed by em-ploying numerical models of laser-tissue interactions on the basis of directlymeasured surface-temperature data [21,22]. The combination of experimentaland simulated data made it possible to characterize surface and bulk tissueheating, thus allowing for a prediction of the temperature rise within theirradiated tissues.

Photochemical welding of tissues has also been investigated as an alterna-tive method for tissue repair without direct use of heat. This technique utilizeschemical cross-linking agents applied to the cut that, when light-activated,produce covalent cross-links between collagen fibres of the native tissue struc-ture. Agents used for photochemical welding include 1,8-naphthalimide [23],Rose Bengal, riboflavin-5-phosphate, fluorescein, and methylene blue [24].Studies of photochemical tissue bonding have been conducted for articularcartilage bonding [23], cornea repair [24], skin graft adhesion [25] and forrepairing severed tendons [26].

As far as the mechanism of tissue welding by means of laser light is con-cerned, a first distinction must be made between: (1) laser tissue welding withor without the addition of exogenous chromophores, (2) laser tissue soldering,and (3) photochemical tissue bonding. For all three cases, the exact under-lying mechanism is not fully understood. Conversely, several hypotheses doexist that are based on a few electron and optical microscopy observationsand on some in vitro studies.

It is widely accepted that laser tissue welding is primarily a thermal process[27]. Thermal modifications of biological components within the connectivetissue have been mainly monitored by means of microscopy. The microscopicdata reported in the literature on laser-welded tissues can be schematicallysplit into two groups on the basis of different modifications of the collagenmatrix observed upon laser welding.

In certain studies, a full homogenization of the tissue was observed, inwhich the loose structure of the collagen fibrils was lost following laser welding


Fig. 15.2. “Hard” laser welding approach resulting in a complete homogenization ofthe tissue. The extracellular matrix is schematically represented by an array of blackdots simulating cross-sectioned collagen fibrils embedded in the ground substance,as in a typical connective tissue (left). Laser irradiation leads to hardly recognizablecollagen structures completely (C) or partially (P) coagulated (right). Temperaturesabove 75◦C are usually recorded in these cases. The mechanism proposed relies onthe adhesion upon cooling between proteins degradated by laser heat, which, actingas microsolders, seal the wound

[4,10,28,29] (Fig. 15.2). In this context, fibrils fused together and morpholog-ically altered revealed a complete denaturation of the collagen matrix [3, 30].As a result of laser irradiation, cell membranes were also disrupted causingleakage of the cellular proteins. In these cases, operative temperatures of over75◦C were induced at the welded area [31]. The wound sealing mechanism hasbeen attributed to the photocoagulation of collagen and other intracellularproteins, which act like micro-solders or endogenous (biological) glue, whichform new molecular bonds upon cooling [28].

In other studies, less severe modifications of the collagen matrix havebeen observed (Fig. 15.3). In these cases, the collagen fibrils were still recog-nizable and not or only partially swollen [3, 32, 33]. Operative temperaturesaround 60–65◦C were generated at the welded site inducing no (or only par-tial) denaturation of fibrillar collagen [33–35]. In water-bath experiments, op-posed tendon specimens were thermally bonded, obtaining maximum tensilestrength at similar temperature values [36, 37]. The most common interpre-tation of the working mechanism is an unravelling of the collagen triple-helixfollowed by “interdigitation” between fibers upon cooling, with generation ofnew bonds [9,32]. Some researchers have suggested the formation of a certaintype of noncovalent bonding between collagen strands on both sides of theweld [38], while others have found that new covalent cross-links were createdupon laser irradiation [39]. A secondary role of fibrillar type I collagen in thelaser welding mechanism has been pointed out in some recent studies [33,40],suggesting the involvement of some other extracellular matrix components,and being in agreement with earlier studies on welded tissue extracts ana-lyzed by gel-electrophoresis [34,41].

280 R. Pini et al.

Fig. 15.3. “Mild” laser welding approach resulting in a partial modification of thetissue morphology. Native connective tissue (left) subjected to laser irradiation ischaracterized by collagen fibrils still recognizable and not or only partially swollen(right). Temperature values in the range 60–70◦C are usually induced at the weldsite. The welding mechanism hypothesized is based on “interdigitation” upon coolingbetween collagen fibres unraveled by laser heat

Photothermal soldering relies on the coagulation of a protein solder bymeans of a laser-induced temperature increase in the tissue. Upon cooling,noncovalent interactions between the solder and the collagen matrix withinthe tissue are supposed to be responsible for the strength of the weld. Evi-dence of albumin intertwining within the collagen matrix was found duringscanning electron microscopy analyzes of specimens irradiated at tempera-tures above 70◦C [17]. Such a threshold value is fully in agreement with thethreshold temperature of albumin coagulation (around 65◦C), as reported inseveral spectroscopic and calorimetric studies. Evidence of extracellular ma-trix infiltration of solder within the tissue was also found by using standardhistological analysis [10,42].

In photochemical tissue bonding photosensitive dyes applied to the woundedges behave as reactive species when irradiated by laser light. They react withpotential electron donors and acceptors such as amino acids (e.g. tryptophan,tyrosine, cysteine) of proteins. Strong covalent bonds are produced betweenthe approximated surfaces of the wound, forming instantaneous protein cross-links [26]. The formation of cross-links in collagen type I molecules by means ofphotochemical activation has been confirmed by using gel electrophoresis [43].

Here as follows, we will review some of the main surgical applications oflaser tissue welding, in particular with regard to the fields of ophthalmologyand microvascular surgery, in which the clinical use of this technique seemscloser to being accepted.


15.2 Laser Welding in Ophthalmology

The first attempt to join biological tissues were proposed in ophthalmology, forthe treatment of retinal, corneal, and scleral samples [44–46]. The first studieswere unsuccessful, resulting in no tissue fusion, while with the improvementof laser technique several research groups adopted different approaches toinduce welding of scleral and corneal tissue. Retina fusion was achieved byinducing photocoagulation of the tissue, while the technique used to treatother ocular tissues is based on a soft thermal treatment, properly definedas laser welding. Successful experimental studies of laser-induced suturing ofocular tissues on animal models have been reported since 1992 by differentauthors [35, 42, 47–51] on the basis of the use of near and far-infrared lasers,directly absorbed by the water content of the cornea. Various laser types withwavelengths exhibiting high optical absorption in water have been used, suchas CO2 (emitting at 10.6 µm) [42, 50, 51], Erbium:YAG (1455 nm) [35], anddiode lasers (1,900 nm) [48,49]. The main problem with such laser wavelengthsis that, without an adequate control of the laser dosimetry, the direct absorp-tion of laser light in a short penetration depth of the tissue outer portioncaused a high temperature rise at the irradiated surface, followed by colla-gen shrinkage and denaturation; on the contrary, the deeper layers are hardlyheated at all, resulting in a weak bonding, as the full thickness of the tissue isnot involved in the welding process. Improved results in tissue welding wereobserved by using exogenous chromophores to absorb laser light, sometimesin association with protein solders. Addition of highly absorbing dyes allowedfusion of wounds at lower irradiation fluences, thus avoiding excessive thermaldamage to surrounding tissues. In fact, the usage of a chromophore was foundto induce a controllable temperature rise only in the area where it had beenpreviously applied, resulting in a selective thermal effect.

Recently a new approach was proposed and tested by some of us for theclosure of corneal wounds [8, 52–54] and for the treatment of anterior lenscapsule bags [55]. It is based on the use of a near-infrared diode laser, in asso-ciation with the topical application of a water solution of ICG. The weldingprocedure, which was optimized so that it could be used in ophthalmic surgeryapplications, has been proposed as a valid alternative to, or as a supportingtool for, the traditional suturing technique used for the closure of cornealwounds, such as in cataract surgery, penetrating keratoplasty (i.e., transplantof the cornea), and in the treatment of accidental corneal perforations. It canalso be used for closure of the lens capsule (to repair capsular breaks causedby accidental traumas or produced intraoperatively), as well as to provideclosure of the capsulorhexis in lens refilling procedures.

15.2.1 Laser Welding of the Cornea

The cornea is an avascularised connective tissue on the outer surface of the eye,and forms the outer shell of the eyeball together with the sclera. It acts as one

282 R. Pini et al.

of the main refractive components, assuring good vision with its clarity andshape. It also acts as a mechanical barrier and as a biological defence system.The cornea is composed of different layers: epithelium, stroma, and endothe-lium are the principal ones, proceeding from the external surface toward theinner part of the eye. More than 90% of the cornea is stroma, and consistsof extracellular matrices (mainly type I collagen and glycosaminoglycans),keratocytes, and nerve fibers. The collagen is regularly arranged in fibers,thus contributing to corneal transparency; the collagen fibers are organizedin lamellae, i.e., in planes running parallel to the corneal surface. These par-ticular structures and architecture confer unique properties to corneal tissue,e.g., in the reaction process to external injuries, such as incidental traumas,surgical incisions, or ulcers. The injured tissues heal by repair and do notrecover the “normal” configuration, because of an induced disorganization inthe ordered array of the collagen fibers. This results in an opaque scar withless tensile strength than that of an unwounded cornea, and in a subsequentimpairment of the main corneal functions. Moreover, the healing of cornealstroma is slower than that of other connective tissues because of the lack ofblood vessels. Clinical changes in scar formation may, in fact, be detectedyears after surgery has taken place [56].

For all these reasons, the characteristics of laser welding procedures may bevery useful in practical surgery, offering the possibility of avoiding many post-operative complications. The immediate watertight closure of wound edgesprovides protection from external inflammation, and may prevent endoph-thalmitis, which sometimes occurs after cataract surgery. The position of theapposed margins has been found to be stable in time, thus assuring optimalresults in terms of postoperative induced astigmatism after cataract [54] andkeratoplasty surgery. The absence or the reduction in the number of stitchesdoes not induce foreign body reaction, thus improving the healing process.Histological analyzes performed on animals and morphological observationson treated patients have shown that, in a laser-welded wound, tissue regainsan architecture similar to that of the intact tissue, thus supporting its mainfunctions (clarity and good mechanical load resistance).

The technique recently proposed for welding corneal tissue [7, 8, 52–54]has been tested and optimized on animal models. Experimental analyzes werefirst performed ex vivo on pig eyes; the healing process was then studied invivo in rabbits. When a satisfactory result was achieved, it was proposedclinically and is currently being used for the closure of corneal tissue afterpenetrating keratoplasty, supporting the application of eight stitches insteadof a continuous suturing. The main advantages of this clinical practice arean increased patient comfort during the healing process and a reduction ofhospitalization costs.

The technique is based on the use of near-infrared continuous-wave AlGaAsdiode laser radiation at 810 nm, in association with the topical application of asterile water solution (10% weight/weight) of ICG to the corneal wound to berepaired. This dye is characterized by high optical absorption around 800 nm


[57], while the stroma is almost transparent at this wavelength. ICG is afrequently used ophthalmic dye, with a history of safety in humans. It is beingused increasingly as an intraocular tissue stain in cataract and vitreoretinalsurgery, as well as in staining of the retinal internal-limiting membrane [58–60].Furthermore, ICG is commonly used as a chromophore in laser welding orlaser soldering [1], to induce differential absorption between the dyed regionand the surrounding tissue. Photothermal activation of stromal collagen isthus induced by laser radiation only in the presence of ICG, resulting in aselective welding effect, which produces an immediate sealing of the woundedges and good mechanical strength. In addition, with the use of ICG, very lowlaser power is required (below 100 mW), and this generally means much saferoperation with respect to the use of other laser types without the associationof dyes.

The procedure used to weld human corneal tissue is as follows: the chro-mophore solution is placed inside the corneal cut, using an anterior chambercannula, in an attempt to stain the walls of the cut in depth. A bubble of airis injected into the anterior chamber prior to the application of the stainingsolution, so as to avoid perfusion of the dye. A few minutes after application,the solution is washed out with abundant water. The stained walls of thecut appear greenish, indicating that the concentration of ICG absorbed bythe stroma is much lower than that of the applied solution. Lastly, the wholelength of the cut is subjected to laser treatment. Laser energy is transferredto the tissue in a noncontact configuration, through a 300-µm core diameterfibre optic terminating in a hand piece, which enables easy handling undera surgical microscope. A typical value of the laser energy density is about13 W cm−2 in humans, which results in a good welding effect. During irradia-tion, the fiber tip is kept at a working distance of about 1 mm, and at a smallangle with respect to the corneal surface (side irradiation technique). This par-ticular position provides in-depth homogenous irradiation of the wound andprevents from accidental irradiation of deeper ocular structures. The fiber tipis continuously moved over the tissue to be welded, with an overall laser irra-diation time of about 120 s for a 25-mm cut length (the typical perimeter ofa transplanted corneal button).

Experimental studies on laser-induced heating effects on ocular tissueswere performed both ex vivo on animal models [21] and in vivo during surgery.An IR thermo-camera provided information on the heating of the externalsurface of the irradiated tissues and on heat confinement, during the treat-ment. A partial differential equation modeling of the process enabled us toinvestigate temperature dynamics inside the tissue. From this study, it waspossible to point out that the optimal welding temperature is about 60◦Cinside the treated wound, i.e., in correspondence with a thermally inducedphase transition of stromal collagen. The heating effect was found to be se-lectively localized within the cut, with no heat damage to the adjacent tis-sue, and to induce a controlled welding of the stromal collagen. After laserwelding, collagen fibrils appeared differently oriented among themselves in

284 R. Pini et al.

Fig. 15.4. Transmission electron microscopy images of the fibrillar arrangementobserved in a control (left) and in a laser-welded (right) corneal stroma. Differentlyoriented and interwoven fibrils across the cut are visible at the weld site upon laserirradiation (×13,500)

comparison with those of the untreated samples, but with similar mean fibrildiameters [33] (Fig. 15.4). Moreover, thanks to the characteristics of the ICGsolution, it seems that laser welding is a self-terminating process, which thusavoids accidental excessive temperature rises inside the tissue.

A follow up study on animal models [7] and clinical application of thetechnique provided evidence that laser-welded tissues exhibits good adhesionand good mechanical resistance (Fig. 15.5). A thorough study of the healingprocess – on the basis of morphological observations, standard histology, andmultispectral imaging auto-fluorescence microscopy (MIAM) [61] – proved ex-perimentally that this takes place in shorter time and with lower inflammatoryreaction, when compared with conventionally sutured wounds. Objective ob-servations 2 weeks after surgery showed a good morphology of laser-treatedcorneas, with almost restored cuts, generally characterized by better adhesionand less edematous appearance compared with sutured ones. These featureswere confirmed by means of histological examinations of rabbit corneas, whichrevealed a well-developed repair process involving the epithelium, which al-most regained its physiological continuity and thickness and a partially reor-ganized architecture of the stroma. Histological analyzes on longer follow-uptimes indicated that the healing of laser-welded wounds was completed inabout 30–60 days, while in sutured wounds the healing process was still inprogress. This result is particularly important as far as corneal tissue is con-cerned, as it typically requires much longer times to be repaired than do othertypes of tissue. Furthermore, the restored tissue regains a stromal architecturethat is very close to the native one, which is crucial to regaining of correctvision.


Fig. 15.5. Left : histological section of a rabbit cornea after laser welding on post-operative day 30 (Hematoxillin and Eosin, ×50); the architecture of the cornearegained an almost physiological appearance (the original cut is indicated by thedashed line). Right : histological section of a rabbit cornea after conventional staysuturing on postoperative day 30; the cut is still clearly detectable and large lacunaeare evident in the corneal stroma

15.2.2 Combing Femtosecond Laser Microsculpturingof the Cornea with Laser Welding

In recent decades, femtosecond (FS) lasers have been developed, tested, andoptimized for miniinvasive intratissue surgery and cell manipulation [62–64].In particular, eye tissue characteristics are well suited to FS lasers applica-tions, as they provide precise intrastromal cuts that are the basis of the newcorneal and refractive surgery techniques. With the term FS lasers, we indi-cate lasers that emit pulses in the near-infrared spectral region with durationsranging between a few femtoseconds and hundreds of femtoseconds. With theuse of these ultra-short pulses, it is possible to induce photo-disruption insidesemitransparent and transparent media, such as corneal tissue. To induce thiseffect, the laser beam is focused in a focal region of a few micrometers in diam-eter. High intensities ∼1013 W cm−2 are achieved, which induce multiphotonnonlinear absorption, followed by plasma formation. The laser-induced plasmaexpands rapidly with high pressure (GPa), and this is followed by developmentand further collapse of cavitation bubbles, accompanied by formation of de-structive shock waves. To obtain a highly-localized destructive effect withoutsignificant collateral damage, small bubble diameters and low photomechani-cal effects in the surroundings are required. These features are satisfied in FSlaser-induced photo-disruption, first because initiation of nonlinear absorp-tion of laser radiation requires tight focusing of the light, which assures effectconfinement in the central portion of the focus volume (where really high

286 R. Pini et al.

intensities are achieved). Second, the multiphoton absorption coefficient hasa high value when compared with the linear absorption value: the disruptiveeffect may be achieved with very low intensity levels (µJ cm−2 or nJ cm−2).These nonlinear effects have been used to ablate and to modify corneal tissue,with very high spatial precision and minimal side effects.

Currently, FS laser cutting cells are characterized by high numerical aper-tures (NA) (>0.9 in water and glass) that make possible submicrometricalprecision. Long series of pulses from FS lasers at 80 MHz repetition rate im-ply accumulative effects, which may cause tissue ablation at pulse energy be-low the optical breakdown threshold in presence of low density plasmas [64].The advantages are that nonlinear propagation effects are reduced, highly lo-calized energy deposition occurs, and subsequently nanosurgery on a cellularand subcellular level is possible. Moreover, the optical breakdown thresholdweakly depends on the target absorption coefficient. Thus, any arbitrary cel-lular structure may be manipulated. For these reasons, near-infrared FS lasershave been considered the innovation for overcoming problems associated withthe use of UV nanosecond lasers in refractive surgery: UV mutation effectson cells, low light penetration depth, collateral damage outside the focal vol-ume, the risk of photon-damage to living cells due to absorption, and prob-able induction of oxidative stress leading to apoptosis. Moreover, a focusednanosecond-pulsed laser beam can cause thermal damage and denaturationaround the laser focus.

Typically, the systems proposed for performing corneal manipulationagainst refractive problems are based on mode-locked diode-pumped FSNd:Glass lasers providing pulses of 500–800 fs duration and a few µJ energy,at repetition rates of some tens of kilohertz. Each individual laser pulse is fo-cused on a specific location inside the cornea, which is fully transparent at thelaser wavelength. A micro-plasma is created, and this generates a microcav-itation bubble of 5–15 µm in diameter, which separates the corneal lamellae.Thus, a resection plane can be created by delivering, in a prescribed pattern,thousands of laser pulses connected together. The cut can be performed withmicrometrical precision at different depths inside the stroma, thus allowingfor corneal flaps with a preset, constant thickness. Laser pulses can also bestacked on the top of each other, to create a vertical or angled cleavage planeto precisely sculpture the border of the lamellar flap. The possibility of per-forming the same resection procedure on the donor cornea as well as on thepatient’s recipient eye, allows to match the transplanted flap precisely withthe recipient corneal bed.

By exploiting previous clinical experiences in diode laser welding of cornealwounds, some of us have recently designed a new laser-assisted techniquefor lamellar keratoplasty (i.e., a corneal transplant involving replacement ofonly the anterior corneal stroma). It is performed by using a FS laser toprepare donor button and recipient corneal bed, and then suturing the edgesof the wound by means of diode laser-induced corneal welding, without theapplication of conventional suture material. This minimally invasive procedure


Fig. 15.6. Aspect of a human cornea affected by leucoma (left): before surgery, and(right): three days after corneal transplant, performed according to the so-calledALSL-LK, which combines the use of a femtosecond laser for corneal sculpturingand a diode laser for corneal welding

was called “all-laser” sutureless lamellar keratoplasty (ALSL-LK) (Fig. 15.6).Intraoperative observations and follow-up results for up to 6 months indicatedthe formation of a smooth stromal interface, the total absence of edema and/orinflammation, and a reduction in postoperative astigmatism, when comparedwith conventional suturing procedures [65,66].

15.2.3 Laser Closure of Capsular Tissue

The lens, or crystalline lens, is a transparent, biconvex structure in the eye:after the cornea, it is the second refractive component in the eye. The lens isflexible and its curvature is controlled by ciliary muscles through the zonules.It is included within the capsular bag, maintained by the zonules of Zinn,which are filament structures connected to the ciliary muscles fulfilling lensaccommodation (i.e., focusing of light rays into the retina in order to assuregood vision). This capsule is a very thin (about 10 µm thick [67]), transpar-ent acellular membrane that maintains the shape of the lens. This tissue is acollagenous meshwork mainly composed of type IV collagen and other non-collagenous components such as laminin and fibronectin. Type I and type IIIcollagen are also present [68]. The function of the lens capsule is primarilymechanical: in the accommodation process it has load-transmitting function.With ageing, the lens loses its ability to accommodate, thus requiring cataractsurgery, which consists of replacing the native lens with a nonaccommodat-ing plastic prosthetic one. The ultimate goal of this surgery is ideally therestoration of the accommodative function by refilling the capsular bag withan artificial polymer [69], after the endocapsular aspiration of nuclear and cor-tical material. This technique may become a viable lens-replacing procedure,

288 R. Pini et al.

as soon as experimental tests are able to prove preservation of capsular me-chanical functions and clarity of refilled lens. Moreover, it would be importantfor the feasibility of this technique, to demonstrate that a biocompatible valveon the anterior lens capsule tissue could be set up to facilitate lens-refillingoperations.

To improve cataract surgery, thus providing a surgical solution to presby-opia, some of us proposed a solution for performing a flap valve with the useof a patch of capsular tissue obtained from a donor lens, to be laser-weldedonto the recipient capsule. The procedure may also be used to repair acciden-tal traumas, such as capsular breaks or perforations during intraocular lensesimplantation [70]. Because of its particular fragility and elasticity, it is quiteimpossible to suture capsular tissue using standard techniques; however, atpresent there are no alternative methods. Laser welding could be used to ac-complish this goal. The study is in progress, and preliminary evidence of thefeasibility of this technique has recently been obtained [55,70].

Experimental tests were carried out ex vivo, on freshly-enucleated porcineeyes (Fig. 15.7). Closure tests were performed by means of patches of donorcapsulae (mean diameter: 3 mm). The inner side of the patch was stained withan ICG-saturated solution in sterile water (7% weight/weight). The stainingsolution was left in place for 5 min. The sample was then washed with abun-dant water, to remove any excess of ICG. The stained patch was then appliedto the anterior lens capsule, through a previously performed corneal incision.

Fig. 15.7. An ICG-stained capsular patch was welded onto the anterior lens capsuleof a pig eye. Laser spots are clearly evident at the periphery of the patch


The stained inner side of the patches were positioned on the exterior surfaceof the recipient capsule, so as to maintain the original orientation and curva-ture. This procedure facilitated adhesion between the tissues to be welded. Thepatch was then irradiated along its external perimeter by means of contiguouslaser spots emitted by a 200-µm-core fiber, whose tip was gently pressed ontothe patch surface (contact welding technique). Exposure times were found tobe critical to avoid heat damage. Continuous wave irradiation, which is typi-cally employed in other laser welding applications, was unsuitable, while pulsesaround 100 ms (with energies of 30–50 mJ) provided the best results. Oncewelded, the capsular patch showed good adhesion to the recipient anteriorcapsular surface. Preliminary biomechanical tests performed on laser-weldedanterior capsule flaps showed that the load resistance of welded specimens wascomparable to that of healthy tissues. Standard histology analysis indicatedgood adhesion between the apposed samples and thermal damage localized inthe treated area.

15.3 Applications in Microvascular Surgery

Microvascular anastomosis is a surgical technique for the connection of twosmall-calibre blood vessels (both arteries and veins) with typical diametersof a few hundreds of micrometers. It is commonly used in various surgicalfields, such as plastic and reconstructive surgery to restore traumatized orthrombotic vessels, as well as in neurosurgery in the treatment of cerebralischemia, vascular malformations, or skull base tumors [71,72]. In this regard,conventional suturing methods are associated with various degrees of vascularwall damage, which can ultimately predispose to thrombosis and occlusion atthe anastomotic site [73]. To minimize vascular wall damage and improve long-term patency, various alternative nonsuture methods have been investigatedexperimentally and, in some cases, in the clinical practice [74–76].

Laser welding of arteries was reported in 1979 by Jain and Gorisch [77] toperform vascular anastomosis by the use of a Nd:YAG laser. Then, other lasers(e.g., argon [78] and CO2 [79]) were used experimentally with questionable re-sults. Further improvements came with the introduction of low-energy diodelasers in association with exogenous chromophores and solders [80, 81]. Morerecently, in vitro and in vivo acute studies have better defined some technicalaspects of end-to-end arterial laser welding [82,83]. Despite the large numberof experimental studies reported in the literature, very few have reached theclinical phase, mainly due to the lack of clear evidence of the advantages of-fered by laser-assisted suturing (when compared with conventional methods),and of reproducibility of results.

Experimental studies on diode laser-assisted end-to-end microvascularanastomosis (LAMA) in association with ICG topical application in femoralarteries and veins of rats were reported by some of us since 1993 [80, 81]. Inthe design and development phases of these studies [80], the number of suture

290 R. Pini et al.

stays supporting the anastomosis was progressively reduced, to minimize for-eign body reaction, and eventually to achieve simpler and faster operation.In conventional suturing procedures of microvascular anastomosis, a numberof suture stays from 8 to 12 are applied to support the anastomosis. On theother hand, laser-assisted procedures to weld the vessel edges were first accom-plished with the support of four conventional stays; then successful LAMAswere performed with only two sutures stays, and lastly with no permanentstays at all. These results demonstrated experimentally that diode laser weld-ing alone could withstand both blood pressure and vessel tensile strength,providing mechanical and functional support to the anastomosis. Moreover,the most significant result of these studies emerged from histological exami-nation on longer follow-up times (up to three months), by comparison of thehealing of laser-welded and conventionally sutured vessels. Specimens fromboth arteries and veins yielded unquestionable evidence of faster and superiorhealing induced by the laser treatment, very similar to a “restitutio ad inte-grum” of vessel structures (Fig. 15.8). The repair mechanism of vessel wallsin LAMAs seemed significantly favored by reduction or absence of suture ma-terial, which limited the occurrence of foreign body reaction, granulomas andinflammation, and ultimately resulted in a more effective and faster restora-tion of the architecture of the vessel walls.

In a more recent study, diode laser welding was experimentally tested toperform end-to-side anastomosis in carotid bypass surgery [11]. This appli-cation may be particularly important in neurosurgery in order to minimizethe occlusion time of the carotid artery during the application of a venousbypass graft, thus reducing the risk of post-operative brain damages. Withregards to technical aspects of the laser-assisted welding procedure, the study

Fig. 15.8. Left : aspect of femorary artery of a Wistar rat after end-to-end LAMAperformed by staining the vessel edges with ICG and then welding them by using alow power diode laser emitting at 810 nm, without conventional stitches. Right : his-tological section of the artery wall 3 months after the surgery showing the completerestoration of the wall architecture (Wiegert elastica-Van Gieson stain, ×250)


demonstrated the feasibility of laser-assisted end-to-side anastomosis in bypasssurgery, which – to our knowledge – had never been tested before in vivo andpresents peculiar technical features. In fact, beside a more complex geometryof the junction, it requires the accomplishment of an effective welding betweenthe artery and the vein graft, which exhibit different wall structures. More-over, surgical advantages were observed in laser-assisted when compared withconventionally sutured anastomoses, such as a simplification of the procedure,a reduction or suppression of bleeding, and a shortening of the operative time,which may be potentially reduced by up to a factor of three. Again, histolog-ical, ultrastructural and immunohistochemical analyzes confirmed the occur-rence of lesser inflammation and of better preservation of the endothelium andof the inner wall structures of both artery and vein in laser-treated segments.This is expected to reduce the occurrence of thrombosis and favor an optimalrestoration process.

15.4 Potentials in Other Surgical Fields

15.4.1 Laser Welding of the Gastrointestinal Tract

Repair of the gastrointestinal tract by means of laser welding has been found tobe a technique that is much easier, faster, and yielding better healing responseand no stone formation, when compared with suturing closure techniques. Thefirst published study regarding laser bowel welding was reported in 1986 bySauer et al. They used a CO2 laser to repair longitudinal transmural incisionsin an otherwise-intact rabbit ileum, producing strong hermetic closures [84].The same authors later proposed the use of a biocompatible, water-soluble, in-traluminal stent in conjunction with India ink as an exogenous chromophore,to perform a suture-free end-to-end small bowel anastomosis [85]. In 1994Rabau et al. described the healing process of CO2 laser intestinal weldingin a rat model, which evidenced a higher probability of dehiscence in thefirst 10 postoperative days compared with control sutured repairs [86]. More-over, comparable healing response among argon and Ho:YAG laser-weldedand control-sutured anastomoses have been found by using a temperature-controlled laser system [87]. Oz et al. investigated the feasibility of using apulsed THC:YAG and a cw argon laser for the welding of biliary tissue. Theyfound better histological response and higher pressure prior to breaking in thelatter case [30]. Lastly, laser soldering with ICG-doped liquid albumin-solderin conjunction with a diode laser was successfully evaluated for the purposeof sealing liver injuries with minimal heat damage [88].

15.4.2 Laser Welding in Gynaecology

Laser welding was experimented in 1978 for reconstructive surgery in thefallopian tubes, using a CO2 laser with good results [89]. Contrasting results

292 R. Pini et al.

appeared in the 1980s up until the study by Vilos et al., in which a series ofintramural-isthmic fallopian-tube anastomoses were performed with a patencyrate of 100% [90]. The main problem claimed in these studies appeared tobe the nonuniform heating of the tissue. Improved results were obtained byintroduction of a low-power diode laser in conjunction with a protein solderin a rabbit model [91]. In this study, no thermal damage was detected, whilea significant reduction in the operative time was evidenced.

15.4.3 Laser Welding in Neurosurgery

The major advantage of using laser welding on nerves is the much shorter time-consumption for operation when compared with conventional microsuturingtechniques. However, as high power densities are required to coapt the nerveedges, thermal damage in the epineurium is easily induced [92]. In contrast,low power does not result in sufficient tensile strength. To improve nervewelding, usage of sealing material was proposed. The most popular soldersare dye-enhanced protein solders, which were employed either with CO2 ordiode laser [28,93]. The bonding rate and the functional recovery of the nervessubject to laser soldering are superior to those of manual suturing techniques,while the risk of inflammation and foreign body reaction are minimized [93].

15.4.4 Laser Welding in Orthopaedic Surgery

Other investigations of the laser tissue welding technique have been carriedout in the case of orthopaedic surgery. The welding of tendinous tissue wasinvestigated using a Nd:YAG laser alone [94] or in conjunction with an albu-min solder, and an argon laser in conjunction with a fluorescein-dye-dopedalbumin solder [95]. These studies pointed out the inability of the laser tech-nique to produce a weld of sufficient strength to withstand the significanttensile loads, which tendons are subjected to from immediately after the re-pair. Laser welding of meniscus tears using an argon laser in conjunction witha fibrinogen solder gave similar results [16].

15.4.5 Laser Welding of the Skin

The principal advantage claimed for the laser welding of skin is a superior cos-metic result. However, control of the temperature enhancement at the weldsite is challenging. It is nearly impossible to obtain full-thickness welds andlimit thermal denaturation laterally around the weld site in the epidermisand papillary dermis. Although thermal denaturation of tissue is necessaryin order to produce a strong weld, excessive thermal damage may result inscarring and dehiscence [96]. The use of chromophores has been proposed tominimize unwanted collateral thermal injury ensuring selective laser absorp-tion at the same time [4, 96]. The application of a variety of solders has also


been investigated for improving the strength of cutaneous repairs [19,97]. Tominimize the effects of collateral thermal damage dynamic cryogen coolingwas successfully employed [98]. A more recent proposal was the use of goldnanoparticles as exogenous absorbers, which allowed for the application oflight sources undergoing minimal absorption from tissue components, therebyminimizing the damage to surrounding tissues [99]. Previously, the ability oflaser welding to accelerate and improve the skin wound healing process wasconvincingly demonstrated [100]. However, it should be emphasized that –until now – clinical work on laser skin welding has never been reported.

15.4.6 Laser Welding in Urology

Urological surgery requires watertight closures, because of the continuous flowof urine. As urine lacks the clotting features of blood, producing effectiveclosures of the urinary tract using traditional suturing techniques tends to betechnically demanding and time-consuming. Moreover, suture material itselfcan induce the formation of stones. Conversely, laser welding can provideimmediate watertight, non-lithogenic anastomoses, yielding a tensile strengththat is superior to that of conventional closure techniques [5]. Laser tissuewelding has been tested on several tissues of the genitourinary tract and fordifferent applications, including urologic applications, which gave the bestclinical achievements.

Laser vasal anastomosis using CO2 and noncontact Nd:YAG laser was ini-tially investigated in several experimental studies, followed by clinical trialswith good postoperative results [101]. Afterwards, laser soldering with ICG-doped albumin in conjunction with an 808-nm diode laser was investigated forhypospadia repair on 138 children [18]. Results were compared with conven-tional suturing. The study emphasized the occurrence of fewer postsurgicalcomplications in the laser group compared with the sutured one, and an easieroperation in the laser set-up.

15.5 Perspectives of Nanostructured Chromophoresfor Laser Welding

Laser welding of biological tissues has received substantial momentum fromcoupling with exogenous chromophores with enhanced absorbance in the near-infrared, applied topically at the edges of the wounds prior to irradiation. Asmentioned earlier, suitable exogenous chromophores absorb efficiently and se-lectively the near-infrared light from a laser, which immediately translatesinto well-localized hyperthermia and an overall decrease of the power thresh-olds required to achieve closure of wounds. This in turn minimizes collateralthermal damage to healthy tissues, which has ultimately backed the emer-gence of laser-welding as a minimally invasive and convenient alternative totraditional suturing or grafting. The ultimate exogenous chromophore should

294 R. Pini et al.

display high absorption coefficient in the near-infrared and enable high local-ization of power deposition, ideally down to the scale of individual biologicalstructures. Further desirable features include good chemical and thermal sta-bility and high photo-bleaching threshold.

Conventional chromophores of common use in laser-welding are organicmolecules such as ICG. These have given outstanding experimental and clini-cal achievements in a number of medical fields, and that in spite of relativelypoor performances with respect to the aforementioned criteria [102,103]. Theabsorption efficiency and photo-stability of organic molecules are limited.Their optical properties depend strongly on biochemical environment andtemperature, and generally deteriorate rapidly with time [104]. The rangeof chemical functionalities accessible is narrow, which is incompatible withflexible and selective targeting of distinct biological structures. Overcomingof these limitations would represent a real breakthrough in the practice oflaser-welding.

Possible ways forward are disclosed by the advent of nanotechnology, asa powerful paradigm to develop new functionalities, by manipulation of self-organization processes at the nanoscale. Here, we mention the introductionof a new class of nanostructured chromophores, which is attracting muchattention in view of many applications, including the laser-welding of bio-logical tissues: colloidal gold nanoparticles (nano-gold). Whereas the opti-cal response of organic molecules stems from electronic transitions betweenmolecular states, light absorption, and scattering in nano-gold originates fromexcitation of collective oscillations of mobile electrons, i.e., surface plasmonresonances [102]. This translates into molar extinction coefficients higher by4–5 orders of magnitude with respect to those of organic chromophores [102],enhanced thermal stability and photo-bleaching threshold, lower dependenceon surrounding chemical environment (although surface plasmons shift withvariations in dielectric constant and electron donating/withdrawing tendencyof embedding tissues [105,106]). Inspiring perspectives arise from the surfacechemistry of nano-gold. As a traditional material for implants, gold is be-lieved to ensure good biocompatibility [107], which is a critical prerequisitein front of clinical applications. The possibility of flexible conjugation of goldsurfaces with biochemical functionalities opens a wealth of novel opportuni-ties [108], such as selective targeting against desired and well-defined biologicalstructures. In summary nano-gold may become the ideal substitute of organicchromophores.

The utilization of nano-gold dates back to the ancient Romans, whenemployed for decorative purposes in the staining of glass artifacts (e.g., theLycurgus cup). Synthesis of stable aqueous colloidal preparations of nano-goldwas first achieved by M. Faraday toward the 1850s by use of phosphorous toreduce a solution of gold chloride. Subsequent developments led to a numberof variants especially based on reduction of chloroauric acid in sodium cit-rate (Turkevich method) [109, 110]. Conventional nano-gold is composed ofspherical nanoparticles of variable and controllable radii [110]. Their optical


properties are well-understood in the framework of classical Mie theory. Theplasmon resonance of spherical nano-gold in aqueous environment is foundwithin 500–600 nm, almost independent of geometrical volume [102].

Because of absorption in the visible, conventional nano-gold is not regardedas a candidate ideal chromophore for laser-welding of biological tissues. Lo-calization of power deposition requires preferential use of near-infrared radia-tion. Theoretical calculations based on different approaches (including Ganstheory [102], dipolar approximations [102, 105], or hybridisation of Mie reso-nances [111]) agree on the possibility to steer the absorption of nano-gold towell within the near-infrared by introduction of nonspherical morphologies.The experimental synthesis of gold nanoparticles with unconventional shapeis a very active field of research [112]. By intrusive modification of existingprocedures for spherical nano-gold, a number of nonspherical gold nanopar-ticles have been demonstrated, including dielectric-core/metal-shell silica orgold-sulphide/gold nano-shells [113–115], complex hollow shells as gold nano-cages [116], or high aspect ratio gold nano-rods [106,117,118]. Tuning of sizeand shape of these nanoparticles allows for tuning of plasmon resonancesin good agreement with theoretical calculations. In particular, absorptionwithin the near-infrared range of interest is becoming a mature achievement.Near-infrared-light irradiation of biological media dispersed with nonspheri-cal nano-gold was proven to result in selective, controllable, and significantheating [119,120].

Photo-activated nonspherical nano-gold holds the promise of manifold ap-plications in the emerging field of nano-medicine. Proposals of extreme interestare e.g., in the treatment of tumors by selective ablation of individual malig-nant cells [120–122]. In our context, the potential of silica/gold nano-shells inthe laser welding of tissues has recently been demonstrated in combinationwith an albumin solder [99]. Absorption of 820 nm diode laser radiation by alow concentration of nano-shells was shown to induce successfully the coagu-lation of albumin proteins and the ensuing soldering of muscles ex vivo andof skin in vivo (rat model). Preliminary results are very promising. However,much progress is still required. We estimate that future innovation will takespecial advantage of the adaptable functionalization of nano-gold. This mayfor example enable the selective targeting of individual biological structures,which may in turn result in the engineering of tissue power absorption pro-files with resolution in the nano-range. This is a completely novel and powerfulperspective in the laser-welding of biological tissues.

Currently, the replacement of traditional organic molecules with nanos-tructured chromophores is an inspiring possibility, which is yet far from theclinical application. As a very recent concept and technology, there exists atpresent basically no experimental evidence of the superiority of these newmaterials in the welding of tissues. Additional aspects of practical concerninclude their biocompatibility and overall sustainability (e.g., economical). Inshort, nano-medicine is a broad and thriving context, offering a wealth of

296 R. Pini et al.

novel opportunities. This is true also in the laser-welding of tissues, which isyet a poorly explored frontier. We foresee exciting evolution in the very nextfuture, possibly along the guidelines sketched in this section.

References

1. K.M. McNally, in Biomedical Photonics Handbook, ed. by T. Vo-Dihn (CRCPress, Boca Raton, 2003), p. 1

2. L.S. Bass, M.R. Treat, Lasers Surg. Med. 17, 315 (1995)3. G.E. Kopchok, R.A. White, G.H. White, et al., Lasers Surg. Med. 8, 584 (1988)4. S.D. DeCoste, W. Farinelli, T. Flotte, et al., Lasers Surg. Med. 12, 25 (1992)5. D.P. Poppas, D.S. Scherr, Haemophilia 4, 456 (1998)6. S.G. Brooks, S. Ashley, H. Wright, et al., Lasers Med. Sci. 6, 399 (1991)7. F. Rossi, R. Pini, L. Menabuoni, et al., J. Biomed. Opt. 10, article 024004

(2005)8. R. Pini, L. Menabuoni, L. Starnotti, Proc. SPIE 4244, 266 (2001)9. R.A. White, G.E. Kopchok, C.E. Donayre, et al., Lasers Surg. Med. 8, 83

(1988)10. B. Ott, B.J. Zuger, D. Erni, et al., Lasers Med. Sci. 16, 260 (2001)11. A. Puca, A. Albanese, G. Esposito, et al., Neurosurgery 59, 1286 (2006)12. C. Chiarugi, L. Martini, L. Borgognoni, et al., Proc. SPIE 2623, 407 (1996)13. Z. Wang, M.M. Pankratov, L.L. Gleich, et al., Arch. Otolaryngol. Head Neck

Surg. 121, 773 (1995)14. S. Wang, P.E. Grubbs, S. Basu, et al., Microsurgery 9, 10 (1988)15. D.F. Cikrit, M.C. Dalsing, T.S. Weinstein, et al., Lasers Surg. Med. 10, 584

(1990)16. S.K. Forman, M.C. Oz, J.F. Lontz, et al., Clin. Orthop. Relat. Res. 310, 37

(1995)17. K.M. McNally, B.S. Sorg, A.J. Welch, et al., Phys. Med. Biol. 44, 983 (1999)18. A.J. Kirsch, C.S. Cooper, J. Gatti, et al., J. Urol. 165, 574 (2001)19. B.S. Sorg, A.J. Welch, Lasers Surg. Med. 31, 339 (2002)20. D.P. Poppas, J.M. Massicotte, R.B. Stewart, et al., Lasers Surg. Med. 19, 360

(1996)21. F. Rossi, R. Pini, L. Menabuoni, J. Biomed. Opt. 12, 014031 (2007)22. W. Small, D.J. Maitland, N.J. Heredia, et al., J. Clin. Laser Med. Surg. 15, 3

(1997)23. M.M. Judy, R.W. Jackson, H.R. Nosir, et al., Proc. SPIE 2970, 257 (1997)24. L. Mulroy, J. Kim, I. Wu, et al., Invest. Ophthalmol. Vis. Sci. 41, 3335 (2000)25. B.P. Chan, I.E. Kochevar, R.W. Redmond, J. Surg. Res. 108, 77 (2002)26. B.P. Chan, C. Amann, A.N. Yaroslavsky, et al., J. Surg. Res. 124, 274 (2005)27. S.A. Prahl, S.D. Pearson, Proc. SPIE 2975, 245 (1997)28. T. Menovsky, J.F. Beek, M.J.C. van Gemert, Lasers Surg. Med. 19, 152 (1996)29. J. Tang, G. Godlewski, S. Rouy, et al., Lasers Surg. Med. 21, 438 (1997)30. M.C. Oz, L.S. Bass, H.W. Popp, et al., Lasers Surg. Med. 9, 248 (1989)31. V.L. Martinot, S.R. Mordon, V.A. Mitchell, et al., Lasers Surg. Med. 15, 168

(1994)32. R. Schober, F. Ulrich, T. Sander, et al., Science 232, 1421 (1986)


33. P. Matteini, F. Rossi, L. Menabuoni, et al., Proc. SPIE 6426, article no. 642614(2007)

34. L.W. Murray, L. Su, G.E. Kopchok, et al., Lasers Surg. Med. 9, 490 (1989)35. H.E. Savage, R.K. Halder, U. Kartazayeu, et al., Lasers Surg. Med. 35, 293

(2004)36. G.M. Lemole, R.R. Anderson, S. DeCoste, Proc. SPIE 1422, 116 (1991)37. P.J. Drew, A. Watkins, A.D. McGregor, et al., Lasers Med. Sci. 16, 291 (2001)38. L.S. Bass, N. Moazami, J. Pocsidio, et al., Lasers Surg. Med. 12, 500 (1992)39. W. Small, P.M. Celliers, G.E. Kopchok, et al., Lasers Med. Sci. 13, 98 (1998)40. M. Constantinescu, A. Alfieri, G. Mihalache, et al., Lasers Med. Sci. 22, 10

(2007)41. C.R. Guthrie, L.W. Murray, G.E. Kopchok, et al., J. Invest. Surg. 4, 3 (1991)42. E. Strassmann, N. Loya, D.D. Gaton, et al., Proc. SPIE 4244, 253 (2001)43. M.M. Judy, L. Fuh, J.L. Matthews, et al., Proc. SPIE 2128, 506 (1994)44. H.C. Zweng, M. Flocks, Tran. Am. Acad. Ophthalmol. Otolaryngol. 71, 39

(1967)45. R.H. Keates, S.N. Levy, S. Fried, et al., J. Cataract Refract. Surg. 13, 290

(1987)46. R.P. Gailitis, K.P. Thompson, Q.S. Ren, et al., Refract. Corneal Surg. 6, 430

(1990)47. N.L. Burnstein, J.M. Williams, M.J. Nowicki, et al., Arch. Ophthalmol. 110,

12 (1992)48. T.J. Desmettre, S.R. Mordon, V. Mitchell, Proc. SPIE 2623, 372 (1996)49. G. Trabucchi, P.G. Gobbi, R. Brancato, et al., Proc. SPIE 2623, 380 (1996)50. A. Barak, O. Eyal, M. Rosner, et al., Surv. Ophthalmol. 42, S77 (1997)51. E. Strassmann, N. Loya, D.D. Gaton, et al., Proc. SPIE 4609, 222 (2002)52. L. Menabuoni, B. Dragoni, R. Pini, Proc. SPIE 2922, 449 (1996)53. L. Menabuoni, F. Mincione, B. Dragoni, et al., Proc. SPIE 3195, 25 (1998)54. L. Menabuoni, R. Pini, F. Rossi, et al., J. Cataract Refract. Surg. (in press)55. R. Pini, F. Rossi, L. Menabuoni, et al., Proc. SPIE 6138, 307 (2006)56. G.S. Schultz, in Cornea. Fundamentals of Cornea and External Disease, vol.

1, ed. by J.H. Krachmer, M.J. Mannis, E.J. Holland (Mosby, St. Louis, MO,1997) p. 183

57. M.L.J. Landsman, G. Kwant, G.A. Mook, et al., J. Appl. Physiol. 40, 575(1976)

58. L.A. Yannuzzi, M.D. Ober, J.S. Slakter, et al., Am. J. Ophthalmol. 137, 511(2004)

59. G.P. Holley, A. Alam, A. Kiri, et al., J. Cataract Refract. Surg. 28, 1027 (2002)60. T. John, J. Cataract Refract. Surg. 29, 437 (2003)61. R. Pini, V. Basile, S. Ambrosini, et al., Proc. SPIE 6138, 404 (2006)62. M. Han, G. Giese, L. Zickler, et al., Optics Exp. 12, 4275 (2004)63. K. Koenig, O. Krauss, I. Riemann, Optics Exp. 10, 171 (2002)64. A. Vogel, J. Noack, G. Httman, et al., Appl. Phys. B 81, 1015 (2005)65. L. Menabuoni, R. Pini, M. Fantozzi, et al., Invest. Ophthalmol. Vis. Sci. 47,

E (2006)66. R. Pini, L. Menabuoni, I. Lenzetti, et al., VII International Symposium on

Ocular Trauma, Rome (2006)67. N.M. Ziebarth, F. Manns, S.R. Uhlhorn, et al., Invest. Ophthalmol. Vis. Sci.

46, 1690 (2005)

298 R. Pini et al.

68. C.C. Danielsen, Exp. Eye Res. 79, 343 (2004)69. E. Haefliger, J-M. Parel, Exp. Eye Res. 10, 550 (1994)70. R. Pini, F. Rossi, L. Menabuoni, et al., Opthalmic Surg. Laser Imaging

(in press)71. M.T. Lawton, M.G. Hamilton, J.J. Morcos, et al., Neurosurgery 38, 83 (1996)72. L.N. Sekhar, C. Kalavakonda, Neurosurgery 50, 321 (2002)73. C.J. Zeebregts, R.H. Heijmen, J.J. van den Dungen, et al., Br. J. Surg. 90, 261

(2003)74. I.A. Aksik, R.P. Kikut, D.L. Apshkalne, Microsurgery 7, 2 (1986)75. E.S. Ang, K.C. Tan, L.H. Tan, et al., J. Reconstr. Microsurg. 17, 193 (2001)76. D.P. Falconer, T.W. Lewis, E.G. Lamprecht, et al., J. Reconstr. Microsurg. 6,

215 (1990)77. K.K. Jain, W. Gorisch, Surgery 85, 684 (1979)78. O.M. Gomes, R. Macruz, E. Armelin, et al., Tex. Heart Inst. J. 10, 145 (1983)79. M.R. Quigley, J.E. Bailes, H.C. Kwaan, et al., Microsurgery 6, 229 (1985)80. R. Gelli, R. Pini, F. Toncelli, et al., J. Reconstr. Microsurg. 13, 199 (1997)81. U.M. Reali, R. Gelli, V. Giannotti, et al., J. Reconstr. Microsurg. 9, 203 (1993)82. A. Lauto, A.H. Hamawy, A.B. Phillips, et al., Lasers Surg. Med. 28, 50 (2001)83. A.B. Phillips, B.Y. Ginsburg, S.J. Shin, et al., Lasers Surg. Med. 24, 264 (1999)84. J.S. Sauer, D.W. Rogers, J.R. Hinshaw, Lasers Surg. Med. 6, 106 (1986)85. J.S. Sauer, J.R. Hinshaw, K.P. McGuire, Lasers Surg. Med. 9, 70 (1989)86. M.Y. Rabau, I. Wasserman, S. Shoshan, Lasers Surg. Med. 14, 13 (1994)87. I. Cilesiz, S. Thomsen, A.J. Welch, et al., Lasers Surg. Med. 21, 278 (1997)88. Y. Wadia, H. Xie, M. Kajitani, J. Trauma 51, 51 (2001)89. L. von Klitzing, R. Grosspietzsch, F. Klink, et al., Fortschr. Med. 96, 357

(1978)90. G.A. Vilos, Fertil. Steril. 56, 571 (1991)91. L.W. Kao, H.R. Giles, J. Reprod. Med. 40, 585 (1995)92. H. Maragh, R.S. Hawn, J.D. Gould, et al., J. Reconstr. Microsurg. 4, 189 (1988)93. A. Lauto, R. Trickett, R. Malik, et al., Lasers Surg. Med. 21, 134 (1997)94. J.D. Burt, M. Siddins, W.A. Morrison, Plast. Reconstr. Surg. 108, 688 (2001)95. F.X. Kilkelly, T.J. Choma, N. Popovic, et al., Lasers Surg. Med. 19, 487 (1996)96. N.M. Fried, J.T.J. Walsh, Lasers Surg. Med. 27, 55 (2000)97. D. Simhon, T. Brosh, M. Halpern, et al., Lasers Surg. Med. 35, 1 (2004)98. N.M. Fried, J.T.J. Walsh, Phys. Med. Biol. 45, 753 (2000)99. A.M. Gobin, D.P. O’Neal, D.M. Watkins, et al., Lasers Surg. Med. 37, 123

(2005)100. A. Capon, S. Mordon, Am. J. Clin. Dermatol. 4, 1 (2003)101. P.T. Gilbert, R. Beckert, Lasers Surg. Med. 9, 42 (1989)102. P.K. Jain, K.S. Lee, I.H. El-Sayed, et al., J. Phys. Chem. B 110, 7238 (2006)103. F. Wang, W.B. Tan, Y. Zhang, et al., Nanotechnology 17, R1 (2006)104. V. Saxena, M. Sadoqi, J. Shao, Int. J. Pharm. 308, 200 (2006)105. M.M. Miller, A.A. Lazarides, J. Opt. A: Pure Appl. Opt. 8, S239 (2006)106. P. Mulvaney, J. Perez-Juste, M. Giersig, et al., Plasmonics 1, 61 (2006)107. R. Shukla, V. Bansal, M. Chaudhary, et al., Langmuir 21, 10644 (2005)108. M.C. Daniel, D. Astruc, Chem. Rev. 104, 293 (2004)109. L.M. Liz-Marzan, Materialstoday 7(2), 26 (2004)110. J. Kimling, M. Maier, B. Okenve, et al., J. Phys. Chem. B 110, 15700 (2006)111. E. Prodan, C. Radloff, N.J. Halas, et al., Science 302, 419 (2003)


112. M. Hu, J. Chen, Z.Y. Li, et al., Chem. Soc. Rev. 35, 1084 (2006)113. S. Kalele, S.W. Gosavi, J. Urban, et al., Curr. Sci. 91, 1038 (2006)114. H.S. Zhou, I. Honma, H. Komiyama, et al., Phys. Rev. B 50, 12052 (1994)115. L.R. Hirsch, A.M. Gobin, A.R. Lowery, et al., Ann. Biomed. Eng. 34, 15 (2006)116. Y. Sun, Y. Xia, Science 298, 2176 (2002)117. J. Perez-Juste, I. Pastoriza-Santos, L.M. Liz-Marzan, et al., Coord. Chem. Rev.

249, 1870 (2005)118. N.R. Jana, L. Gearheart, C.J. Murphy, J. Phys. Chem. B 105, 4065 (2001)119. C.H. Chou, C.D. Chen, C.R.C. Wang, J. Phys. Chem. B 109, 11135 (2005)120. L.R. Hirsch, R.J. Stafford, J.A. Bankson, et al., Proc. Natl. Acad. Sci. 100,

13549 (2003)121. J.Y. Chen, F. Saeki, B.J. Wiley, et al., Nano Lett. 5, 473 (2005)122. X. Huang, I.H. El-Sayed, W. Qian, et al., J. Am. Chem. Soc. 128, 2115 (2006)

16

Photobiology of the Skin

A.P. Pentland

16.1 Basics of Skin Structure: Cell Types,Skin Structures, and Their Function

Everyone is familiar with the skin because it is the organ that is primarilyresponsible for the appearance of human beings. In addition to this impor-tant role, skin also has a myriad of important maintenance functions. Thespecialized cells of the skin work together to form the protective and sup-portive structures necessary for our bodily integrity [1]. The main structuresof the skin include the epidermis, which is the most superficial layer of theskin, and forms its outer barrier [2] (Fig. 16.1). The epidermis is supportedunderneath by the dermis, the portion of the skin that is responsible for thephysical integrity of the skin, and gives it the capacity to undergo stretchingand flexing and resist trauma [3–5]. Interspersed in this tough matrix is theskin’s vascular network and the nerves that permit the transmission of signalsindicating pain or itch. Lastly, there are many adnexae situated in skin, suchas the hair, nails, sebaceous, and sweat glands. Collectively, the key functionof these structures is to keep toxic environmental influences out, and preventloss of bodily fluids [1]. Some specific aspects of these functions are:

• Temperature regulation; sweating and vasodilation for heat loss [6]• Physical integrity; dermal toughness and elasticity [4, 5]• Maintenance of water balance; providing an intact evaporative barrier [2]• Microorganism defense; skin has active innate and acquired immunity [7]• Appearance [8]• Use of light to synthesize Vitamin D [9]• Protection from hazardous UV radiation [10]

The cell types present in the epidermis are keratinocytes, melanocytes, andlangerhans cells. Keratinocytes constitute approximately 90–95% of the cellsin epidermis, and are responsible for its structure [1]. Keratinocytes are pro-duced in the innermost layer of the epidermis adjacent to the dermis. They

302 A.P. Pentland

Skin cell Types

- Keratinocytes - Melanocytes

- Vascular cells- Mast cells

- Fibroblasts

- Adnexal cells

Stratum Corneum

Stratum granulosum

Stratum Spinosum

Stratum Basale

Epiderm

is

Derm

is

- Langerhans cells

Fig. 16.1. Skin is a multilayered organ. Epidermis comprises the most superficiallayer. The cells of the epidermis divide in the basal layer, then differentiate, movingupward to the skin surface to slough. The dermis underlies the epidermis, providingstructural support for blood vessels and adnexae, as well as the epidermis

then differentiate, flattening and progressing toward the surface of the skin asthe process of differentiation proceeds, until they are no longer viable. Dur-ing this differentiation process, keratinocytes synthesize and secrete an arrayof lipids intercellularly in organelles termed lamellar bodies. As differentia-tion is completed, these organelles are secreted into the extracellular spacewhere their lipid contents spread and disperse among the cells of the stratumcorneum, creating a “brick and mortar” motif that is extraordinarily imper-meable [2,11]. The barrier is resistant to moisture, invasion of microorganisms,and even oxygen. The constant sloughing of fully differentiated epidermal cellsalso is an adaptive strategy that prevents microorganism invasion and pro-vides a constant new supply of skin as we wash and scrub the skin surface inthe course of work and play. Keratinocytes are also capable of innate immuneresponses, secreting inflammatory cytokines and chemotactic mediators thathelp to repel invading organisms [7], and are responsible for the synthesis ofthe vitamin D precursor, Vitamin D3 [9, 12].

A second important cell type located in the epidermis is the melanocyte,which represents ∼4% of the cells in the epidermis [13]. It is responsible forour skin color because of its capacity to synthesize melanin granules and se-crete them to be taken up by keratinocytes [14,15]. The melanin produced bymelanocytes is the chief photoprotective substance within the skin, as it is ca-pable of absorbing light throughout the visible and ultraviolet light spectrum.One melanocyte provides melanin for about 36 keratinocytes (Fig. 16.2).

In addition to these two key cell types, the epidermis also contains ∼4%Langerhans cells, which are the resident immune cell of the epidermis. Whenmicroorganisms do successfully penetrate the epidermal stratum corneum

16 Photobiology of the Skin 303

Fig. 16.2. The melanocyte is responsible for the synthesis of melanin. Melanocytesreside in the basal layer of the epidermis. They extend dendrites to many nearby ker-atinocytes. Melanin is delivered to keratinocytes by transfer of pigment-containingmelanosomes to these nearby keratinocytes, where it is situated above the nucleus.Melanin is responsible for skin color

barrier, it is the Langerhans cell that identifies and transports microorgan-ism proteins to regional lymph nodes so an adaptive, long-term, and specificimmune response can be initiated.

The dermis underlies the epidermis, creating its physical support structure.It contains fibroblasts, vascular cells, adnexal cells, mast cells, and dermal den-drocytes [4,5]. Fibroblasts are responsible for the synthesis of dermal proteins.The most abundant of these is type 1 collagen, which is present in the dermisas a highly cross-linked protein network. In addition to collagen, fibroblastssynthesize and secrete elastic fibrils, which impart resiliency and elasticity toskin and comprise ∼1–2% of skin protein. Mature elastic fibers in the dermis

304 A.P. Pentland

are composed of an amorphous elastin core surrounded by microfibrils, alsoproduced by fibroblasts. Interestingly, most synthesis of elastin occurs in thefirst trimester, so that as chronic UV exposure degrades this protein, our skinloses its stretchiness, causing wrinkles [8, 16].

In addition to fibroblasts, the dermis contains a vascular network thatprovides nourishment to the skin, and which can dilate and contract as neededto adjust the core temperature [6]. Mast cells are located at the intersectionsof the small vessels in the dermis where they can readily detect minor traumaand invading microorganisms. The mast cell is the key cell responsible for hivesobserved in allergic reactions [17]. The mast cell also contributes to the skin’sinnate immune defenses. When stimulated physically or by proteins presenton invading organisms, it releases granules that contain vasodilator substancesand mediators that can activate the immune system. These substances “kickstart” the host immune response.

Interposed between the epidermis and the dermis is a thin layer of elegantlyassembled attachment proteins, termed the basement membrane [18, 19].There are approximately 15 different proteins present in the basement mem-brane zone, which work in concert to permit the rapidly proliferating epidermisto maintain its attachment to the underlying dermis.

16.2 Effects of Light Exposure on Skin

As in any other system, light interacting with the epidermis is reflected, re-mitted, scattered, or absorbed. Only absorbed solar radiation has a biologicaleffect on the skin. The most bioactive wavelengths in skin are those in theultraviolet range, from 280 to 400 nm [10]. Although some work has beendone to demonstrate that 670 nm light may improve wound repair [20, 21],most literature on the interaction of light with skin focuses on wavelengthsin the ultraviolet range, because of their capacity to promote synthesis ofvitamin D [9], cause photoaging and skin cancer [22], as well as their thera-peutic utility.

Radiation in the ultraviolet range is divided into two types, on the basisof the associated biology. Wavelengths from 280 to 320 nm are termed UVBlight, while those from 320 to 400 nm are termed UVA light. UVB light isprimarily responsible for initiation and promotion of skin cancer, and vitaminD synthesis [10, 12]. UVA wavelengths also contribute to skin cancer, butare much less effective. Their primary impact is on photoaging. Both UVAand UVB wavelengths stimulate melanin synthesis by melanocytes to protectagainst the unwanted effects of light [23]. The biological effects of these twotypes of light are dictated by the cellular structures that absorb light. DNAabsorbs light readily in the UVB range, and therefore UVB light is far moremutagenic than UVA light [24].

When skin is exposed to UV light, it produces both acute and chroniceffects. In the acute response, observed 4–48 h after exposure to UV light,


skin can develop redness due to vasodilation, swell, become painful, or evenblister if the dose of light is sufficiently large. Somewhat after these changesoccur, in 2–7 days, tanning and stratum corneum thickening occur. Thoseindividuals who have very little capacity to tan will not produce much extramelanin after UV exposure, but will still produce thickening of the stratumcorneum [9]. The time course of the acute changes resulting from UV exposureis also dependent on the dose of light. Doses of light that are in large excessof the photoprotective capacity of an individual’s skin will result in muchmore long-lasting and intense erythema [9]. Thus, the changes produced inskin are very dependent on the genetic capacity for melanin synthesis, whichis termed the skin type [25]. Skin types are rated on a scale of 1 through6 (Fig. 16.3). Individuals with type 1 skin always burn and never tan whenexposed to UV light. Individuals with type 2 skin burn initially, but then cantan modestly after recurrent exposure to light. Those with type 3 skin mayburn initially in the spring, but will reliably tan thereafter. Those with type 4skin have significant coloration without sun exposure, and always tan readily.Those with type 5 skin have deep coloration without sun exposure, but tan-ning is evident after sun exposure. Lastly, individuals with type 6 skin haveconstant very deep pigmentation in which tanning is difficult to detect. Light-related skin cancer is prevalent primarily in skin types 1–3. Over the course ofyears, recurrent exposure of individuals to UV light produces chronic changesin susceptible, light skinned, individuals. Most evident is photoaging. Char-acterized by deep wrinkling, uneven pigmentation, vascular dilatation, andhyperkeratotic changes in epidermis, photoaging is what most people think

Fig. 16.3. Skin types: capacity to tan, burn, and risk of skin cancer

306 A.P. Pentland

of as aging because of the passage of time [8]. However, vigilant protectionagainst exposure to UV light or a naturally dark complexion are highly ef-fective in preventing these skin changes [9]. For those individuals who arelight skinned, chronic exposure to light can also result in immunosuppressionand skin cancer formation [22]. Despite these negative consequences of UVlight exposure, UV light can be useful in the treatment of skin disease. Theresponse of skin to light is shaped by the presence of many chromophoresin skin. These include aromatic amino acids, DNA, NADH, beta-carotene,porphyrins, and most importantly melanin, which has an absorptive rangefrom 280 nm through the whole visible spectrum of light [22, 26]. Althoughabsorption of light by melanin is generally harmless, absorption of light byDNA, amino acids, and membrane lipids produces immediate cellular dam-age. Therefore, intricate defensive mechanisms are in place to assess the extentof UV damage that has been produced by a particular exposure to UV lightand to initiate appropriate cellular responses to it.

Immediately after exposure to UV light, the cell division process is paused(Fig. 16.4). UVA wavelengths produce damaging free radical production, whileboth UVA and UVB can produce photoproducts in DNA [10,27]. The key pro-tein initiating this halt in cell cycling is p53, sometimes called “the guardianof the genome” because of its role in response to UV injury. At about the sametime, a transcriptional activation occurs. This permits induction of DNA ex-cision/repair enzymes so that damaged DNA can be repaired. Efforts at DNArepair are focused mostly on actively transcribed genes. In addition, oxidizedor denatured proteins are degraded, and the synthesis of their replacementsis initiated. Similar mechanisms are carried out to repair damaged lipids.Synthesis of inflammatory cytokines is also initiated in response to light ex-posure, including TNFα, IL-1, INFγ and TGFα, which act to regulate repairresponses and also are immunomodulators [24, 28, 29]. As new antigens andDNA fragments can be produced by exposure of the skin to light, these cy-tokines help to ensure that sun exposure does not result in sun allergy orautoimmune disease. About 72 h after sun exposure, melanin synthesis andincreased differentiation of keratinocytes is prominent [30]. By 7 days afterexposure to light, the repair of DNA and proteins is back to baseline values,and the photoprotective capacity of the exposed epidermis has been increasedas needed. In some instances after UV exposure, the damage to cells is sosevere, that repair is not a viable option. In this case, cellular apotosis isinduced. Cells undergoing apoptosis after UV injury were initially describedas sunburn cells. Apoptotic destruction of injured cells allows their orderlyremoval from injured tissue [31].

The process of repair that occurs after UV exposure is imperfect, mak-ing chronic UV exposure problematic. Mutations in DNA can escape repairprocesses, an error that is key for cancer initiation. Mutations may increaseexpression of genes that support proliferation or block genes that inhibitit. The capacity of UV light to both cause mutations and to stimulate aproliferative response, as described earlier, is the reason that UV light is


Fig. 16.4. After exposure to UVB radiation, a cellular process to assess the responseof epidermis to UV injury is activated. Cell division is halted, the extent of injury isassessed, and repair processes are initiated. Concurrently, those individuals capableof tanning induce melanin synthesis and pigment transfer to keratinocytes. Thosecells that have been severely injured are removed by the process of apoptosis

a complete carcinogen, both initiating and promoting cancerous growth inskin [24]. Clones of these initiated cells proliferate as recurrent UV exposureoccurs over time, increasing the opportunity for a second mutation to occurin a cell, and thereby causing cancer [10]. The added twist that UV light isimmunosuppressive further increases the capacity of UV light to cause cancer,as the immune surveillance capability of irradiated skin is reduced, permit-ting survival of more initiated cells. After UV exposure, Langerhans cells,

308 A.P. Pentland

the resident immune cell in epidermis, are depleted [24]. Immunosuppressivecytokines are released, including IL-10, TNF, αMSH and PGE2. PGE2 alsopowerfully stimulates keratinocyte cell division [32]. Although this immuno-suppression is necessary to prevent sun allergy and autoimmune disease, itdecreases the ability of individuals with poor native photoprotection to en-dure sun exposure over the course of their lifetime without developing skintumors.

The mutations in DNA induced by UV exposure have been well character-ized. Adjacent thymine residues in DNA form cyclobutane pyrimidine dimers.Also formed are pyrimidine–pyrimidone photoproducts, which can isomerizeto a Dewar isomer. Should replication of DNA occur before repair of thesephotoproducts, base substitution and therefore mutation, can occur [10]. Inaddition to this type of DNA damage, UV light can also produce strand breaks.Additionally, UVA light is capable of producing an 8-oxo-guanine photoprod-uct in DNA. Once DNA photoproducts occur, they are removed over time.Those that are most likely to be mutagenic are removed more efficiently thanthose that are not. The removal of photoproducts is increased in areas of ac-tively transcribed DNA. It is also much quicker in newborns than in olderindividuals (Fig. 16.5).

With so much evident harm resulting from exposure of the skin to light,it is not surprising that the skin is well endowed with defenses against UVinjury. In addition to the repair pathways already described, and the broadUV absorbing capacity of melanin, the epidermis also has other substances inabundance that can absorb UV energy harmlessly. Glutathione is present inthe cell cytosol in large amounts, as is urocanic acid, which isomerizes whenabsorbing energy in UV wavelengths. In addition, epidermis contains catalaseand superoxide dismutase to detoxify UV-induced free radicals [9].

Fig. 16.5. UV light is a complete carcinogen, capable of initiating mutations inDNA that also decrease the capability of cells to detect these alterations. It alsopromotes mutated cells to divide, increasing the chance of a second mutation withina cell, and therefore increasing the chance of cancer progression


One good activity that occurs when skin is exposed to UV light is the syn-thesis of Vitamin D [9,12]. Provitamin D (7-dehydrocholesterol) is synthesizedin epidermis. When this compound is exposed to light wavelengths between280 and 315 nm, it is altered to become previtamin D3. This compound isthen isomerized by the warmth of the skin to become Vitamin D3 (cholecal-ciferol). Vitamin D3 is sufficiently soluble that it can then enter circulationto be metabolized to 25(OH)D3 termed calcediol, or 25-hydroxy vitamin D3.This vitamin is important in calcium deposition in bones and calcium home-ostasis in the kidney. UV exposure is not a requirement for adequate suppliesof Vitamin D; however, it is readily obtained from dietary sources.

16.3 Sun Protection and Sunscreens

Skin cancer is the most common form of cancer, with nearly a million new casesoccurring every year. The most dangerous type of skin cancer, melanoma, israpidly rising in incidence, with lifetime risk now approaching 1 in 50. Allforms of skin cancer are linked to sun exposure [22, 33]. It is, therefore, quiteimportant to be cautious about sun exposure, particularly for those with lightcomplexions, and to adopt sun protection practices to minimize skin cancerrisk and also to decrease the effects of photoaging. The simplest and mosteffective method for minimizing sun exposure is to avoid outdoor activitiesduring the period of the day when UVB exposure is most intense, from 10amto 4pm. In addition, when outdoors, good sun protection can be achieved bywearing protective clothing, a hat with a 4′′ brim, and a sunscreen [9].

Some clothing is also now rated by its sun protection factor, so an indi-vidual can determine the degree of protection provided by a garment [34,35].Some garments provide more protection than others. The style of the garmentmatters – it must cover the skin. Two layers are better than one, and newergarments and thicker garments are more UV resistant than older ones. Inter-estingly, even the type of fiber makes a difference. The UV blocking capacityof fibers is as follows: polyester > wool > silk > nylon> cotton and rayon. Ofcourse wet, see-through fabric also does a poorer job filtering the light thandry fabric.

As individuals vary so widely in their sunburn responses to UV light, thecommon method to rate protection is called the sun protection factor, or SPF.The SPF of a sun protection device or lotion is determined empirically. Aninitial exposure of a subject’s skin to graded doses of a UV light source ismade to determine the minimum amount of light that can produce detectableerythema with a distinct border 24 h later. Clearly, the more melanin an in-dividual has in their skin, the larger the dose of light necessary to produceredness. Once the minimal erythema dose (MED) is determined for an in-dividual, the protective measure to be tested is applied to the area to beirradiated and the skin again exposed to light (Fig. 16.6). The capacity of theprotective measure to block erythema production 24 h later is expressed as

310 A.P. Pentland

Fig. 16.6. The sensitivity of an individual to damage from sun exposure is measuredby determining their minimal erythema dose. The least amount of light capable ofproducing a clearly outlined spot of redness, (termed erythema) is called the minimalerythema dose for that individual. The amount of light required to produce erythemavaries nearly 100-fold because of the differences in constitutive skin color

the sun protection factor, or SPF. For instance, if a garment has an SPF of15, then an individual who ordinarily develops erythema after 10 min of lightexposure without protection will now be able to tolerate 150 min of exposurewhen wearing the garment [9].

Use of sunscreens as a method of protection from skin cancer risk hasbeen actively debated in recent years. There is definitive evidence that theiruse protects animals from skin cancer in the laboratory. Similar trials havenot been conducted in humans, but there is strong supportive evidence. Somedebate has arisen about whether sunscreens may interfere with Vitamin Dmetabolism, but it is clear that the amount of Vitamin D needed can bereadily obtained from dietary sources. In addition, very little light is needed toproduce Vitamin D from endogenous sources. It would be the rare individualwho could be sufficiently assiduous in their sun protective measures to produceVitamin D deficiency.

Most currently available sunscreens were initially formulated to protectagainst UVB wavelengths, from 280 to 320 nm. There are two major classesof these compounds: UVB absorbing and physical sunscreens [9]. With recentrapid advances in material science, many new sunscreen compounds are beingcreated and tested. Some of these new sunscreens have already been testedfor safety and met standards for European approval, while some aspects ofsafety for new sunscreen materials are still being evaluated.

Physical sunscreens are effective across a broad range of wavelengths. Theyare made up of particulates that can scatter and reflect UV light. In general,they contain zinc oxide, titanium dioxide, magnesium oxide, talc, or kaolin(up to 25%). Sunscreens in this category provide the best block available in


the UVA range (up to 380 nm), and are also very effective at blocking UVB.Their chief drawback is that when formulated to have a high sun protectionfactor, they are visible as a white substance when applied to skin. Muchwork has been conducted lately to alter their physical properties to decreasescattering such that visibility is decreased. Some of this alteration has beendone by making nanoparticle formulations. Unfortunately, their capacity toprotect against light in the UVA range is decreased when the particle sizeis decreased into the nanometer range. This problem is being addressed bygrouping some nanoparticles in the sunscreen vehicle, and also by coating theparticles with dimethicone or silica. A recent concern has been raised thatthe capacity of nanoparticles to penetrate the skin may be different from thatof larger particles. Because of the character of the stratum corneum, it maybe they cannot penetrate readily into viable tissue in significant quantities.Research is ongoing in this area [9].

UVB absorbing sunscreens are capable of harmlessly absorbing UV light,and are generally degraded over time by their absorption of light. Sunscreensof this type are wavelength selective, unlike the physical sunscreens. Therefore,it is necessary to combine several chemical sunscreens to gain optimal coverageacross the UV spectrum. More consideration is also now being given to group-ing sunscreen ingredients so that their absorption of UV can be transferred toheat, thereby leaving the sunscreen ingredient intact and able to function inits absorbing capacity longer. UVB blockers include paraaminobenzoic acid(up to 15%), Padimate O, a PABA ester (up to 8%), Octocrylene (up to10%), Cinnamates (such as octyl methoxycinnamate, up to 8% ), and Sali-cylates (up to 5%). Some chemical sunscreens are broader in their protectivecapacity, such as the benzophenones and oxybenzone (up to 6%), which areuseful for UVB protection, but also protect against the shorter wavelengthsin the UVA spectrum. Until recently, the only available absorbing sunscreenprotective in most of the short UVA spectrum was Parsol 1789, which isutilized in concentrations up to 3%. New formulations of UV absorbing mole-cules are now becoming available, which will greatly expand the capacity ofsunscreen manufacturers to make a product that protects broadly across UVwavelengths.

16.4 Phototherapy: Use of Light for Treatment for SkinDisease

Despite the problems associated with exposure to UV light, it has been recog-nized for centuries that it is effective in the treatment of skin disease. Diseasessuch as psoriasis, eczema, and vitiligo have been managed with moderate ef-ficacy, using sun exposure for centuries [9]. With the advent of artificial lightsources, it became practical to expand the use of light as a treatment option,as therapy could be administered throughout the year and in cold climates.

312 A.P. Pentland

Three types of phototherapy are in common use today, on the basis ofthe light source in use for treatment. The type of lamp selected for use intreatment depends upon the patient’s diagnosis, the disease location and thepatient’s skin type. All types of phototherapy are less effective in dark-skinnedindividuals because melanin blocks some of the therapeutic light transmission.The most widely available is UVB therapy, using lamps that emit light pri-marily in the wavelengths from 280 to 320 nm. Because of its similarity to solarUVB emission, treatment can be administered daily, taking advantage of thebody’s adaptive responses to UVB light. In use since the early part of the lastcentury, it is a well-characterized treatment method. Because the wavelengthsof light emitted are the same wavelengths responsible for photocarcinogenesis,it is used sparingly. Studies examining whether UVB phototherapy producesincreased cancer risk suggest that the increase is modest, due to physiciansupervision of light treatment and the fact that its use is generally intermit-tent for disease therapy. Tanning also occurs when patients are treated withthis light source, which decreases the efficacy of light treatment. Disease lo-cated deep within the skin is difficult to treat because of absorption of UVBwavelengths primarily in the epidermis [9].

In addition to UVB phototherapy, two other light sources have been in-troduced in the more recent past for phototherapy treatment. The creationof UVA light sources (320–400 nm) triggered an investigation of whether thislight source could be useful for disease located deep in the skin, which is re-sistant to UVB treatment because of poor skin penetrating capacity. To makeUVA wavelengths efficacious as a treatment method, light exposure occursafter patients have taken the photosensitizing drug psoralen. The treatmentmethod is, therefore, termed “PUVA,” meaning psoralen + UVA. When pso-ralen is present in the skin during UVA irradiation, psoralen becomes cross-linked to DNA. This initiates the repair pathways discussed earlier, and hasa beneficial effect on a number of diseases. Because the treatment producesmore inflammatory change than does UVB therapy, it is only given 2–3 timesper week to allow ample time for repair. As this treatment also producescrosslinks in DNA, the risk of photocarcinogenesis is quite high; studies havedemonstrated that the rate of squamous cell skin cancer occurring after PUVAincreased eightfold and that melanoma rates are increased as well. Nonethe-less, PUVA treatment is significantly more efficacious than UVB treatment,and therefore in cases of refractory disease, it is used [9]. The lamps usedin tanning parlors are the same as those used for PUVA treatment, but thepublic does not take a photosensitizing drug prior to their visit!

To try and reduce the undesirable side effects of UVB and PUVA treat-ment, lamps that emit light most intensely at 312 nm have been brought tomarket, termed as narrow-band UV lamps. This light source avoids some ofthe cancer risk of broad-band UVB lamps, and can work well without pro-ducing intense tanning of patients. This form of treatment has been found tobe nearly as effective as PUVA, and is becoming the standard of care in manytreatment centers. The diseases most responsive to phototherapy using any


of the light sources mentioned earlier are psoriasis, atopic dermatitis, vitiligo,refractory itching, some forms of lymphoma, and many more [9].

In summary, the skin is a complex organ with many functions. Althoughwe are constantly reminded of its role in appearance, it has important utilityin protecting us from the environment. There are many special adaptations ofskin that are involved in protecting us from UV light, some of which we areable to exploit to treat disease.

References

1. D. Chu, A. Haake, K. Holbrook, L. Loomis, The Structure and Development ofSkin, 6th edn. (McGraw-Hill, New York, 2003)

2. R.R. Wickett, M.O. Visscher, Am. J. Infect. Control 34, S98 (2006)3. J.M. Waller, H.I. Maibach, Skin Res. Technol. 12, 145 (2006)4. J. Uitto, L. Pulkkinen, M. Chu, Collagen, 6th edn. (McGraw-Hill, New York,

2003)5. J. Uitto, M. Chu, Elastic Fibers, 6th edn. (McGraw-Hill, New York, 2003)6. B. Wenger, Thermoregulation, 6th edn. (McGraw-Hill, New York, 2003)7. G. Stingl, D. Maurer, C. Hauser, K. Wolff, The Skin: An Immunologic Barrier,

6th edn. (McGraw-Hill, New York, 2003)8. M. Yaar, B. Gilchrest, Aging of Skin, 6th edn. (McGraw-Hill, New York, 2003)9. H. Lim, H. Honigsman, J. Hawk, Photodermatology (Informa Healthcare, New

York, 2007)10. D.I. Pattison, M.J. Davies, Experientia Supplementatum 96, 131 (2006)11. P. Elias, K. Feingold, J. Fluhr, Skin as An Organ of Protection, 6th edn.

(McGraw-Hill, New York, 2003)12. R. Ebert, N. Schutze, J. Adamski, F. Jakob, Mol. Cell Endocrinol. 248, 149

(2006)13. R. Halaban, D. Hebert, D. Fisher, Biology of Melanocytes, 6th edn. (McGraw-

Hill, New York, 2003)14. G. Scott, S. Leopardi, S. Printup, N. Malhi, M. Seiberg, R. Lapoint, J. Invest.

Dermatol. 122, 1214 (2004)15. Z. Abdel-Malek, M.C. Scott, I. Suzuki, A. Tada, S. Im, L. Lamoreux, S. Ito, G.

Barsh, V.J. Hearing, Pigment Cell Res. 13(Suppl 8), 156 (2000)16. A. Oba, C. Edwards, Skin Res. Technol. 12, 283 (2006)17. F. Hseih, C. Bingham, K. Austen, The Molecular and Cellular Biology of the

Mast Cell, 6th edn. (McGraw-Hill, New York, 2003)18. T. Masunaga, Connect. Tissue Res. 47, 55 (2006)19. L. Bruckner-Tuderman, Basement Membranes, 6th edn. (McGraw-Hill, New

York, 2003)20. M.T. Wong-Riley, X. Bai, E. Buchmann, H.T. Whelan, NeuroReport 12, 3033

(2001)21. H.T. Whelan, R.L. Smits Jr., E.V. Buchman, N.T. Whelan, S.G. Turner, D.A.

Margolis, V. Cevenini, H. Stinson, R. Ignatius, T. Martin, J. Cwiklinski, A.F.Philippi, W.R. Graf, B. Hodgson, L. Gould, M. Kane, G. Chen, J. Caviness, J.Clin. Laser Med. Surg. 19, 305 (2001)

22. R.N. Saladi, A.N. Persaud, Drugs Today 41, 37 (2005)

314 A.P. Pentland

23. W. Westerhof, O. Estevez-Uscanga, J. Meens, A. Kammeyer, M. Durocq, I.Cario, J. Invest. Dermatol. 94, 812 (1990)

24. M. Kripke, H. Ananthaswamy, Carcinogenesis: Ultravioloet Radiation, 6th edn.(McGraw-Hill, New York, 2003)

25. T.B. Fitzpatrick, Arch. Dermatol. 124, 869 (1988)26. F.R. de Gruijl, H.J. van Kranen, L.H. Mullenders, J. Photochem. Photobiol. B

63, 19 (2001)27. X. Chen, A. Gresham, A. Morrison, A.P. Pentland, Biochim. Biophys. Acta

1299, 23 (1996)28. V.E. Reeve, Methods 28, 20 (2002)29. T. Nakamura, I. Kurimoto, S. Itami, K. Yoshikawa, J.W. Streilein, J. Dermatol.

Sci. 23(Suppl 1), S13 (2000)30. G. Scott, A. Deng, C. Rodriguez-Burford, M. Seiberg, R. Han, L. Babiarz, W.

Grizzle, W. Bell, A. Pentland, J. Invest. Dermatol. 117, 1412 (2001)31. B.J. Nickoloff, J.Z. Qin, V. Chaturvedi, P. Bacon, J. Panella, M.F. Denning, J.

Investig. Dermatol. Symp. Proc. 7, 27 (2002)32. R.L. Konger, R. Malaviya, A.P. Pentland, Biochim. Biophys. Acta 1401, 221

(1998)33. A.C. Geller, G.D. Annas, Semin. Oncol. Nurs. 19, 2 (2003)34. V.E. Reeve, M. Bosnic, D. Domanski, Photochem. Photobiol. 74, 765 (2001)35. D. Domanski, M. Bosnic, V.E. Reeve, Redox Rep. 4, 309 (1999)

17

Advanced Photodynamic Therapy

B.C. Wilson

17.1 Introduction

Photodynamic therapy (PDT) is the use of drugs (photosensitizers) that areactivated by visible or near infrared light to produce specific biological effectsin cells or tissues [1]. The basic steps in a PDT treatment are application of thephotosensitizer (systemically or topically), a time interval to allow for photo-sensitizer accumulation in the target diseased tissue or cells, and illuminationof the target area or volume with light of an appropriate wavelength to activatethe sensitizer. PDT is a highly multidisciplinary topic, involving optical bio-physics and bioengineering, synthetic chemistry, pharmacology, photophysicsand photochemistry, photobiology, and different clinical specialties. The mainemphasis in this chapter is on those aspects of greatest interest to specialistsin biophotonics.

The term “photodynamic” was first coined a century ago with the ob-servation that light and an acridine dye in the presence of oxygen could killmicroorganisms [2]. At around the same time, light therapy was being usedin patients but without any administered photosensitizing agent. The firstreported clinical use of PDT was in 1904 when eosin was applied locally toa tumor on the lip and exposed to light. There was limited follow up to thisearly work, probably due to the lack of potent photosensitizers and of suitablelight generation/delivery technologies. The modern era of PDT started withthe discovery of hematoporphyrin derivative (HpD) in the 1950s/1960s, andthis was first used in patients in the 1970s to treat bladder cancer. Substan-tial preclinical and clinical studies were started in the late 1970s and the firstgovernment approval for PDT was in 1993, with a purified version of HpD(Photofrin R©).

In addition to being further developed since as a treatment for solidtumors and precancerous conditions, PDT is currently approved in variouscountries for nononcological applications [3,4]. These include actinic keratosis(sun-damaged skin) and age-related macular degeneration, in which abnormalblood vessel growth in the retina causes central vision loss, particularly in the

316 B.C. Wilson

elderly. Recently, there has also been a large interest in PDT to treat localizedinfection [5], in part driven by the rise of bacteria and other microorganismsthat are multidrug resistant. For example, PDT using the agent methyleneblue was recently approved for treating gum infection (periodontitis) and isalso approved for treating acne using the agent ALA (see later): in both casesthe sensitizer is applied topically.

These applications exploit the different biological mechanisms that canbe selectively activated, depending on the PDT treatment parameters: cellkilling, blood vessel shut down, and possibly immunological effects [6, 7].

The invention of the laser and optical fibers in the 1960s was also animportant driver for the development of PDT, since these enable light ofadequate intensity to be delivered to almost anywhere in the body and somake the treatment generically applicable. It should be emphasized, however,that PDT remains largely a localized treatment, since the penetration of lightin tissues is limited, so that it is not generally useful for systemic diseases.

17.2 Basic Principles and Features of “Standard PDT”

Before discussing the potential novel approaches to PDT, we will summarizethe “standard” PDT technique that is characterized by the following:

– High doses of photosensitizer and light, given as a single treatment (evenif this is subsequently repeated to achieve complete clinical response).

– Single-photon activation (see Fig. 17.1 and discussion), using continuous-wave (CW) or quasi-CW light sources, and exploiting the Type II pho-toreaction.

– Photosensitizers that are always activatable.– Photosensitizers that comprise organic molecules that are directly acti-

vated by the light.

Figure 17.1 shows the Jablonski, or energy-level, diagram for PDT. Theground state (S0) of the photosensitizer absorbs the photon energy and israised to an electronic excited (S1) state (lifetime approximately nanosec-ond). This can de-excite either nonradiatively by fluorescence emission to S0

or by intersystem crossing to a triplet state (T1). The T1 →S0 transitionis quantum-mechanically forbidden, so that T1 is long-lived (approximatelymicrosecond) and can exchange energy with ground-state molecular oxygen(3O2) to generate singlet-state oxygen, 1O2, which is highly reactive and leadsto oxidative damage to nearby biomolecules (within 10 s of nanometer). Mostoften these are cell membrane components because of the photosensitizer mi-crolocalization. Following this photoreaction, S0 is regenerated, so that thephotosensitizer essentially acts as a catalyst. It may eventually be destroyed(photobleached), either directly by the interaction or indirectly through dam-age due to the 1O2. In addition to the properties common to any therapeuticdrug, photosensitizers should have high 1O2 quantum yield, high absorption

17 Advanced Photodynamic Therapy 317

Fig. 17.1. Jablonski diagram for singlet oxygen generation by PDT (Type II pho-toreaction)

at red/near-infrared wavelength to get good light penetration, high target-to-nontarget uptake ratio in tissues or cells, and low concentration in the skinand eyes to avoid complications from accidental light exposure. Photosensi-tizers that are not water soluble require formulation into liposomes or lipidemulsions in order to achieve efficient delivery in vivo. The absorption spectraof some PDT photosensitizers are shown in Fig. 17.2, together with a plotshowing the general dependence of light penetration in tissues. For example,HPD has its highest absorption in the UVA/blue region and only a very smallred absorption (at about 630 nm). This contrasts with most second-generationphotosensitizers where the molecule is designed to have the largest absorptionpeak above 630 nm where the hemoglobin absorption falls rapidly, but belowabout 800 nm in order to maintain high 1O2 quantum yield and avoid the NIRwater absorption bands.

One agent that is used widely is aminolevulinic acid (ALA). While notitself a photosensitizer, administering ALA to cells leads to synthesis of thephotosensitizer protoporphyrin IX (PpIX) [8]. The advantage is that thereis a degree of intrinsic tissue selectivity that does not just rely on enhancedphotosensitizer uptake. ALA-PDT is used in treating tumors, particularlyearly-stage, and other indications mentioned above. Its fluorescence has alsomade it useful for early cancer/precancer imaging, either for tumor detectionor in guiding tumor surgery, particularly in the brain [9]. As well as selection ofgood photosensitizer properties (and enough oxygen), successful PDT requiresthat enough light must be delivered to the target tissue to generate biologicallyeffective levels of 1O2 (>108 molecules per cell). This can be technologicallychallenging, since the amount of light required is quite high [10], typically>100 J cm−2 of red/NIR light incident on the tissue surface. Since this amountof light has to be delivered in at most tens of minutes for clinical practicality

318 B.C. Wilson

efficiency

Photofrin

400

400

1

5

10

600 800 1000

500 600 700

mTHPCAIPcS4

Wavelength (nm)

Monte Carlo

r [mm]

z [mm]

Wavelength (nm)

effe

ctiv

e p

enet

rati

on

dep

th(m

m)

Useful range formost (non-superficial)

applications

Ab

sorp

tio

n

1ms(1-g) = 1 mm

21

0.1

1

10

100

1000

104

−1

1

0

0

Wcm2

per Wincidentpower

Fig. 17.2. Absorption spectra of some typical photosensitizers (top) and the de-pendence of light penetration in tissues (bottom), showing the range of effectivepenetration (1/e) depth spectra in typical tissues (left) and a representative MonteCarlo calculation of the distribution of light fluence in tissue of given optical ab-sorption and scattering properties (right)

(but see mPDT below), the light sources require typically a few Watts ofoutput power at the right wavelength for the photosensitizer. Suitable lightsources are diode lasers, wavelength-filtered lamps, and light emitting diodearrays (LEDs) [3, 11]. Each has its own advantages and limitations. Overall,lasers are preferred when delivery via single optical fibers is required, forexample, in endoscopy or when treating a larger tissue volume by multipleinterstitial fibers; LED arrays are particularly useful for irradiation of easilyaccessible tissue surfaces. Lamps have a role when treating very large surfaceareas and have the advantage of easy (and low cost) change of wavelength tomatch different photosensitizers.

Different “modes” of target illumination by the light are used: surface, in-terstitial, or intravascular; external, endoscopic, or intraoperative. In all cases,making the total system (light source, power supply, light delivery devices,etc.) ergonomic and economic for specific clinical applications is critical. Forthe highest efficacy, the spatial distribution of the light in the tissue mustbe “matched” as far as possible to the volume and shape or the target tis-sue. This can be very difficult, given (a) the limited penetration of light in


tissue (the 1/e depth of red/NIR light in tissues is typically only a few mil-limeter), (b) the high variability optical scattering and absorption propertiesof tissues, and (c) the limited degree of “tailoring” of the applied light dis-tribution that can be achieved, especially in body locations where access isrestricted. A variety of light-delivery devices has been developed for differentapplications [3, 10]. For example, optical fibers can be modified to produce acylindrical irradiation volume along the last several centimeter at the tip, andrecently it has become possible to make the output distribution deliberatelynonuniform to match the target volume profile. Likewise, various balloon de-vices have been developed to irradiate within body cavities. With multiplefiber delivery, efficient splitting of the output of the light source is needed,and recent developments have included instruments in which nonequal distri-bution among sources can be achieved, again to match the target geometryin individual patients. These different technologies are still evolving as newclinical applications are investigated. A major challenge in some applicationsis to make the light source and delivery much cheaper and simpler to use thanthe established systems where, typically, the source cost is ∼$10 K per Wattand single-use delivery devices are hundreds of dollars.

17.3 Novel PDT Concepts

In this main section we will present several new concepts that are still underdevelopment but which, if eventually adopted into clinical practice, wouldrepresent “paradigm sifts” in the way that PDT is performed, either in termsof the underlying principles used and/or in the technologies involved.

17.3.1 Two-Photon PDT

The photophysical interaction that has been used in essentially all PDT appli-cations to date is that shown in Fig. 17.1 (although, in some cases, there maybe Type 1 contributions to the photobiological effect). Thus, the S0 state ofeach molecule is activated by absorbing a single photon. Two-photon (2γ) ac-tivation, representing a fundamental alternative to this scheme, is illustratedin Fig. 17.3, in two different implementations both involving the absorptionof two photons of light for each activation cycle. Consider Case A (referredto here simply as 2γ activation), where a wavelength is used at which thephotosensitizer has a nonzero 2γ cross-section, σ. The probability of simulta-neous absorption of two photons is then proportional to σI2, where I is theinstantaneous light intensity. Hence, to avoid requiring too much total energy(which could cause thermal damage to the tissue) an ultrashort laser pulse isneeded, typically ∼100 fs. Since the energy required for the S0 →S1 transitionis essentially the same as for 1γ activation, the wavelength of the pulsed lasercan typically be pushed further into the NIR range, thereby increasing thepenetration depth in tissue. This is one of the putative advantages of this ap-proach, although the situation is complex, since it involves a balance between

320 B.C. Wilson

S1 S1

S0

1O2

3O2 S0

T1 T1

Tn

Fig. 17.3. Jablonski diagrams for two different types of 2γ activation involvingeither (a) simultaneous absorption of two photons by S0 (left) or (b) sequentialabsorption of two separate photons by S0 and then T1 (right)

the reduced tissue attenuation and the quadratic dependence on the activa-tion intensity, which falls off faster with depth than the linearly-dependent1γ intensity. The second advantage is that, because of the I2 dependence,focusing the laser beam activates the photosensitizer only within a very small(∼femtoliter) tissue volume. This is the same effect as used in 2γ confocal mi-croscopy, where it offers the analogous advantage of reduced photobleachingof the sample above and below the focal plane.

A potential application of this concept is to treat AMD, which is a majorapplication for (1γ) PDT, involving targeting of the abnormal blood vesselgrowth in the retina. 2γ PDT may be able to achieve vessel closure but min-imize any collateral damage caused by activation of photosensitizer outsidethe blood vessel layer. Recent research indicates that this is possible, in prin-ciple [12], but that there are several substantial technical challenges to beovercome, due primarily to the fact that absorption of two photons by S0 isa very low probability event [13]. As a result, since activation is proportionalto σI2, either σ and/or I must be very high. The former requires the de-velopment of new molecules that have σ values several orders of magnitudehigher than conventional PDT photosensitizers: in turn, this may also requirethe use of a “designer” delivery vehicle, since the pharmacological propertiesmay then be difficult to control (e.g., limited water solubility). In principle,the intensity, I, may be arbitrarily high. However, in practice, this is limitedby several factors: (a) the need to keep the total energy within sub-thermalrange, and thus limiting the pulse energy, which means shorter pulses, (b) thecost of such laser sources, (c) nonlinear processes in the propagation of thelaser pulses in the delivery system and the tissue itself, and (d) the onset ofphotomechanical damage to the tissue even in the absence of photosensitizer,which loses the potential selective advantage of photosensitizer localization.

The practical implementation envisaged for 2γ PDT is to image the regionof abnormal blood vessels, or “feeder” vessels, using a confocal scanning laserophthalmoscope (cslo), as illustrated in Fig. 17.4. This instrument operateson the same principle as confocal microscopy, using a scanning laser beamfocused at a specific plane in the object (in this case the neovascular layer).


Fig. 17.4. Confocal scanning laser ophthalmoscope under development for image-guided 2γ PDT of age related macular degeneration, showing the optical design(top), the system in use, and a typical high-resolution retinal image (normal)

The operator would then define the “target” and the laser beam would beswitched to “treatment mode,” using the high peak-power fs laser. Apart fromthe fs light source itself, there are several optics challenges in implementing2γ PDT for AMD, in particular, the following:

– Efficient coupling of the fs laser source into the cslo is complicated by ef-fects such as pulse dispersion that take place as the laser beam propagatesthrough the optical elements

– Achieving diffraction-limited focusing at the back of the eye is restrictedby the low numerical aperture of the eye and possibly by scattering (e.g.,by cataract) and wavefront distortion (e.g., from astigmatism).

To tackle the second problem, adaptive optics approaches will likely be neededand, in addition, automatic eye tracking will be required to keep the focal spotat the correct location. Both technologies are under development.

Case B, so-called two-photon/two-color activation (2γ/2λ)-PDT, uses twolaser pulses of different wavelength and separated in time. The first pulse

322 B.C. Wilson

generates T1 via the S1 state. A second pulse, at a wavelength that is stronglyabsorbed by T1, then generates higher-order triplet states, Tn. The biologicaleffects are then mediated by the Tn states interacting with nearby biomole-cules. The key point is that this process does not require molecular oxygen,and so may avoid the limitations of Type II photosensitizers, which do requireadequate free oxygen in the target tissue. Oxygen-independent cell killing by2γ/2λ PDT has been demonstrated in cells, although using a photosensitizerthat is poorly suited to clinical use. However, there are major challenges inmoving this concept to the clinic: (a) the photosensitizer must have high ab-sorption in both the S0 and T1 states and these must be at useful wavelengthsfor tissue penetration and (b) a 2γ/2λ laser source is required, preferablywith independent control over the intensity and wavelengths of each pulseand pulse–pulse time delay to match the sensitizer photokinetics. Work is inprogress to address both challenges. For example, Mir et al. [14] are devel-oping phthalocyanine with specific structural modifications to yield optimumabsorption spectra in the ground and excited states, while rapid developmentsin pulsed laser sources, such as fiber lasers, are likely to yield practical lasertechnologies in the next few years. (Note that, in principle, this pathway couldbe activated to operate in CW mode with both wavelengths present simulta-neously, although the photoefficiency would be very low.)

17.3.2 Metronomic PDT

At the other end of the light intensity scale is the concept of metronomic pho-todynamic therapy (mPDT). This is based on the observation that it may bepossible to increase the target-specific cytotoxicity by delivering the photosen-sitizer and the light at very low dose rates over an extended period (hours ordays) rather than applying a single, high dose-rate treatment. An example ofsuch specificity is seen in treating brain tumors [15]. In this case, the clinicalchallenge is that the tumor cells around the edges of a solid tumor infiltratethe normal brain tissue. With standard, high dose-rate treatment, it may notbe possible to kill the tumor cells without also destroying normal brain tis-sue. However, tumor cell-specific programmed cell death (apoptosis) can beachieved by mPDT. This has been demonstrated for ALA. It is not known ifthe mPDT effect is unique to ALA or applies also to other photosensitizers,or whether the mPDT principle can be applied also to other tumor sites ordiseases.

From the biophotonics perspective, a challenge is to develop light sourcesthat could deliver light over a long period and still be practicable, either forclinical use or in preclinical animal models. Figure 17.5 shows an examplebased on high-efficiency coupling of LEDs into optical fibers, with batteryand driver electronics incorporated into a lightweight “backpack.” For clinicaluse, the technology challenges will include (a) how to distribute the light, e.g.,in the brain or other site after surgery to remove the bulk tumor tissue, (b)how to integrate the power supply, light source (diode laser or LED), and the


Fig. 17.5. Fiber-coupled LED sources developed for mPDT use in rodent brain, anda source in use to treat brain tumor in a rat model, with the battery and electronicsin a wearable backpack

light delivery components, (c) how to maintain sterility and biocompatibilityof such an implant and (d) how to provide power, for which one option is tohave an external power supply and remote coupling to the source, as is usedin some other implantable devices.

The fundamental biological issue with mPDT is whether or not it is pos-sible to kill the target cells faster than their proliferation, while avoiding non-apoptotic killing (to avoid inducing an inflammatory response that could causesecondary damage to normal tissues) and avoiding direct PDT damage to nor-mal tissues.

17.3.3 PDT Molecular Beacons

Molecular beacons were introduced several years ago as a way to enhancethe target specificity of molecular imaging agents [16]. In that case the con-cept was to link a fluorescent reporter molecule (fluorophore) with a molecule

324 B.C. Wilson

Triplet energy transfer

Enzyme

hννO2

1O2+

PSQ

hννO2

1O2

PSQ

Enzyme Activation

PS

Linker

5’3’

G -CA -TG -CC -GG -C

Q

1O2


O2

1O2hνννν

GA

GC

G

PS

LinkerStem

Linker

Tumor SpecificmRNA

mRNA

AS-ON

PSPS

Linker

5’3’


Q

1O2

Linker

5’3’


Q

1O2

5’3’


Q

1O2


Q

1O2


QQQhννO2

1O2

AS-ONLoop


O2

1O2hνννν

GA

GC

G

Linker

O2

1O2hννν

GA

GC

G

PS

LinkerStem

Linker

+

0.E+00

1.E+04

2.E+04

3.E+04

4.E+04

5.E+04

6.E+04

BufferAlone

P30 P30+RNA

P30C P30C+RNA

P30C+30min

Tot

al 1 O

2 Lu

min

esce

nce

Fig. 17.6. PDT molecular beacons. showing the concept with enzyme cleavage ofthe linker (top), and using an antisense loop (bottom), with an example of the low1O2 luminescence signal (third bar) compared to controls (first and second bars)that increases with unquenching (fourth bar) and then decreases with inhibition ofthe unquenching (fifth bar)

(quencher) that suppresses the fluorescence when the two are in close prox-imity. The quenching efficiency has a very strong (∼1/r6) dependence on thequencher-fluorophore distance, r. The linker molecule could then be broken by,for example, interaction with an enzyme that is found in high concentrationin the target tissues or cells.

This concept has recently been transferred into PDT using a photosensi-tizer instead of the fluorophore (Fig. 17.6). Proof-of-principle studies of thisconcept have been reported [17], showing, for example, that 1O2 can be gen-erated by light exposure at much higher levels after unquenching than before.Selective toxicity has also been shown, both in vitro and in vivo, where onlycells/tumors that express the specific enzyme used in the beacon are sensitiveto PDT.

There are several possible variations on this idea, in terms of the linkerstructures. One of the most interesting, but challenging in its synthetic chem-istry, is to use a so-called antisense loop, in which the quencher and photo-sensitizer are held in close proximity when the “hairpin” structure is closed,but then the linker opens up and unquenches the photosensitizer when ithybridizes with a specific nucleotide sequence in the target cells. Antisensebeacons have been demonstrated to date in solution. This could have tremen-dous disease specificity, since it is based on highly defined genetic biomarkercharacteristics of, for example, tumor cells.


17.3.4 Nanoparticle-Based PDT

There has been great excitement recently in the potential of nanoparticles(NPs) for a wide range of applications, including medical diagnostics andtherapeutics [18]. This disruptive technology is based on the special optical,electrical, mechanical, or chemical properties that emerge when materials areformulated at the nanoscale (typically, 1–100 nm) compared to the bulk prop-erties. Optical NPs include the following:

– Metal NPs with light absorption and scattering spectra that are highlydependent on the size, shape, and form (e.g., solids vs. shells, spheres vs.rods)

– Quantum dots (Qdots), semiconductor NPs that are fluorescent in thevisible and/or near-infrared, which are very bright, have very low photo-bleaching, have narrow emission spectra with the peak emission increasingwith the Qdot diameter, and have a wide excitation spectrum so that asingle light source can excite many different sizes of Qdots

– NPs that are themselves not optically active but that can carry a “pay-load” of optically active molecules, such as fluorescent dyes or photosen-sitizers

In addition to these specific novel properties, NPs have the advantagethat they can be “decorated” with molecules, such as antibodies or peptidesequences, to target them to specific cells or tissues. The potential of NPs forcancer diagnostics (particularly imaging) and therapeutics has been recentlyreviewed [19]. An important subset of these techniques is photonics-based,including fluorescence imaging and photothermal and photodynamic treat-ments.

For PDT there are several potential and distinct ways to utilize NP char-acteristics, as illustrated in Fig. 17.7, namely (a) as “carriers” of PDT photo-sensitizers, (b) as photosensitizers themselves, or (c) as “energy transducers.”

Fig. 17.7. Schematic of three different uses of nanoparticles for PDT: (a) as targeted“carriers” of the photosensitizer, (b) as direct 1O2 generators, and (c) as energytransducers

326 B.C. Wilson

1O2 luminescence peak in suspension

9000000

8000000

7000000

6000000

5000000

4000000

3000000

2000000

1000000

01200 1220 1240 1260

Wavelength (nm)

Ph

oto

n c

ou

nts

1280 1300 1320 1340

Fig. 17.8. 1O2 signal from porous silicon NPs in water (relative to background):insert shows the ultrastructure of these NPs

For Case A, a variety of NPs can be used that either encapsulate many pho-tosensitizer molecules per NP or have many attached to the surface in orderto deliver a high payload of photosensitizer to the target cells/tissues. Thereare several demonstrations in the literature of this approach. One advantagethat PDT has over, say chemotherapeutic drugs, is that generally the toxicityof most photosensitizers is very low in the absence of light activation, so thatthere is a higher tolerance for nontarget “leakage” in the delivery system.

For Case B, Fig. 17.8 shows an example where singlet oxygen is generatedupon light irradiation of porous silicon NPs, without any added molecularphotosensitizer. This is believed to be due to direct energy transfer to themolecular oxygen in the liquid that permeates the NPs, which have an ex-tremely large surface-to-mass ratio. It has not yet been shown that this canactually be exploited to produce PDT cell killing: these studies are in progress.Potential barriers are getting the NPs into the cells and having the 1O2 dif-fuse from the (inner) NP surface to the biological target. (Note that in thecase of NPs as photosensitizer delivery vehicles, it may not be necessary forthe NPs themselves to penetrate the cell wall if the payload can be releasedwithin the target tissue, such as in the interstitial space, from which it wouldthen be taken up by the cells.) In the third scenario, Case C, Qdots are usedas the primary light absorber. The energy is then transferred to a photosen-sitizer molecule that is conjugated to the Qdot, activating it to the S1 state,as in standard PDT. The generation of 1O2 via Forster Resonant EnergyTransfer has been demonstrated in solution and should work in cells/tissues.


Again, the advantage in principle would be the ability to target to Qdots tothe desired tissues/cells without compromising the optimum photosensitizerproperties. A variation of this approach would be to use it for 2γ PDT, sinceQdots are known to have very high σ values (∼104–105 GM units comparedwith ∼10 GM units for conventional 1γ drugs and ∼103 GM units for the best“designer” 2γ molecules to date).

17.4 PDT Dosimetry Using Photonic Techniques

PDT dosimetry, i.e., the measurement of the “dose” that determines the effi-cacy of the treatment, is complex, due to the following:

(a) The multiple factors involved (photosensitizer, light, oxygen)(b) The heterogeneity of the local photosensitizer concentration, light fluence

rate, and oxygen level in the tissue, which may result in nonuniform de-position of PDT dose throughout the target volume, and

(c) The interdependence and dynamic behavior of the parameters, e.g., 1O2

destroys the photosensitizer molecules (photobleaching), reducing theamount available for activation; the photosensitizer may increase the at-tenuation of light in the tissue, further limiting the penetration depth;if the photosensitizer concentration and the light intensity (fluence rate)are high enough, this can deplete the ground-state oxygen faster than theblood supply can replenish it.

Three main dosimetry approaches have been used to address this complexity,termed explicit, implicit, and direct [20]. In the first, the attempt is to measurethe three basic parameters of light fluence rate, photosensitizer concentration,and molecular oxygen concentration, which are then combined into a predic-tive estimate of the effective PDT dose. There are many technical details ineach of these three measurements [3, 10]. Figure 17.9 illustrates light dosime-try in PDT to destroy prostate cancer by delivering light through multiplediffusing optical fibers to the whole prostate volume.

To optimize the distribution of light throughout the prostate treatmentplanning is used through a series of steps: 3D MRI is done to determine thevolume and shape of the prostate; source fibers are placed at selected posi-tions to cover the prostate volume; the spatial light fluence-rate distributionresulting from delivering selected powers to each of the source fibers is cal-culated; this is compared with the desired light distribution; and the stepsare iterated to optimize the number, length, position, and power from eachsource. This is analogous to treatment planning in radiation therapy (e.g.,brachytherapy using implanted radioactive wires or seeds). For photosensi-tizer dosimetry, the main technique has been to measure the fluorescent bypoint spectroscopy or imaging. However, while it is straightforward to makerelative measurements, it is more difficult to derive absolute photosensitizerconcentration values, since the measured signal depends also on the tissue

328 B.C. Wilson

Fig. 17.9. Example of light dosimetry during PDT of prostate cancer, showingthe patient set-up, the placement of fibers and detector fibers through a template,the multichannel light dosimeter measurements of the local light fluence rate as afunction of time during treatment at three different locations the graph shows: (a)in the prostate, (b) the urethra and (c) the rectum

optical properties. One solution is to measure these properties separately andapply a light propagation model to correct for the attenuation. Alternatively,fiberoptic probes have been developed to reduce this attenuation dependence,for example, by sampling only a very small tissue volume [22].

The measurement of oxygen levels in tissue is well-established, for example,using interstitial microelectrodes or fiberoptic probes. Noninvasive measure-ments are less well developed and do not directly measure the pO2, e.g., diffusereflectance spectroscopy can detect changes in the relative concentration of


Fig. 17.10. Example of treatment planning for PDT of prostate [21], showing theprocess for calculating the 3D light distribution in the tissue (top, two-source ex-ample) and the diffusion-theory equation that is solved on a nonuniform mesh byfinite-element analysis (bottom, six-source example)

hemoglobin and oxyhemoglobin, from which the oxygen saturation, SO2, canbe derived. (This is also used in studies of brain and muscle physiology [23].)

In implicit dosimetry a surrogate measure of the integrated interplayof light, drug, and oxygen is employed, most commonly the photosensi-tizer photobleaching. Referring to Fig. 17.1, if the 1O2 can react with anddisrupt the photosensitizer molecules such that they become nonphotoac-tive/nonfluorescent, then this is related to the 1O2 production. This method istechnically fairly simple and can be applied at multiple locations in the tissues.

330 B.C. Wilson

However, interpreting the data rigorously is not simple, since the photobleach-ing may or may not be 1O2-dependent and there may be non-1O2 dependentcytotoxic pathways if the tissue pO2 is low. Nevertheless, under well-controlledconditions, photobleaching can be a very useful PDT dose metric [24].

Lastly, in direct dosimetry, 1O2 itself is quantified by detecting the1,270 nm luminescence emission from the 1O2 →3O2 transition. The signalis very weak, since 1O2 is so reactive in biological media, with only about 1in 108 molecules decaying by luminescence. However, this has been demon-strated recently both in vitro and in vivo and is highly correlated to the PDTresponse [25–27]. The challenge now is to simplify and reduce the cost andcomplexity of the technology.

17.5 Biophotonic Techniques for Monitoring Responseto PDT

Partly in response to the continuing challenge of PDT dosimetry, but also togain more direct understanding of the biological effects of PDT treatment,there has been ongoing interest in assessing directly the tissue responses fol-lowing or even during PDT. The last, in particular, is also motivated by thepossibility to use this information for feedback control of the treatment whileit is in progress.

A first example that is used for preclinical studies is bioluminescence imag-ing (BLI). For this the cells of interest are transfected with the luciferase(“firefly”) gene, so that, when luciferin is supplied the cells emit visible lightthrough chemiluminescent. Although weak, this can be imaged noninvasivelyusing a high sensitivity photodetector, such as a cooled CCD array. BLI hasbeen used to quantify changes in implanted tumors [28] or the survival of bac-teria [5] treated with PDT in animal models. The method can be extendedby having the luciferase gene only “turned on” when another gene of interest(e.g., a stress-response gene) is up-regulated by the PDT treatment, so thatBLI gives direct in vivo imaging of the response of the target cells at thespecific molecular level.

A second example is optical coherence tomography (OCT), the analogueof high-frequency ultrasound imaging, based on interferometry (Fig. 17.11)(explained in detail in the chapter by S. Boppart). Depth scans (A-scans) areproduced by detecting the interference signal between the reflected light froma given depth in the tissue and a reference beam: high spatial resolution isobtained by using a light source with a very short coherence length. A 2Dimage (B-scan) is then generated by scanning the beam across the tissue sur-face. OCT is approved in ophthalmology and is being widely investigated fora range of other applications, including endoscopic and intravascular imag-ing using fiberoptic implementation. As with ultrasound, OCT can operate inDoppler mode to image blood flow (by measuring the small shift in phase be-tween successive A-scans caused by movement of the red blood cells or othertissue components).


Change in microvascular blood flow index during and post PDT

1

0PRE 1 3 6 12 24h POST

------DURING----

3 5 7 20Hours post PDT treatment

low-coherenceCW optical source

detector

tissue

beamsplitter

referencemirror

tissue

coherence length

structure blood flow

Fig. 17.11. Doppler OCT for monitoring vascular response of tumor tissue to PDTtreatment. (Top) Principle of OCT and example of cross-sectional imaging in vivo.(Bottom) Structural and DOCT images at different time points following PDT treat-ment showing blood vessel shut down (left) and plot of volume-averaged measure ofblood flow in the tissue during and following Photofrin-PDT treatment of melanoma

For monitoring the changes to tissue from PDT, one can look at either thestructural or the Doppler images. The latter have proved to be more usefulto date, since the changes in tissue structure are rather subtle. Figure 17.11also shows examples of the changes in blood flow in individual microvesselsdue to PDT. Such changes can be seen in some cases even during the lighttreatment [29], so that there is the possibility to use the DOCT informationto modify the treatment “on line.” A limitation with OCT is that the depthof imaging is small (∼1–2 mm). Recently, we reported an OCT probe within asmall-diameter needle that can be placed interstitially [30], e.g., at the tumorbase to ensure that adequate PDT dose is delivered.

17.6 Biophotonic Challenges and Opportunitiesin Clinical PDT

We end with a brief overview of some of the major outstanding limitations ofPDT and how these might be overcome, at least in part by applying biopho-tonic science and technologies.

Certainly PDT can be both safe and effective. However, there can be sig-nificant variations in the response. One of the obvious problems is that thereis little or no “tailoring” of the treatment parameters used to the individual

332 B.C. Wilson

patient or lesion. The various dosimetry methods outlined above could im-prove this situation, but they are rarely used in routine clinical practice atthis time, in part because they are still rather complicated to use and inter-pret. A major technical challenge is to devise the next generation of dosimetrydevices that will be simple and fast to use, minimally invasive, inexpensive,and reliable. Until this happens, PDT will continue to evolve largely as anempirical procedure with optimization based on dose ranging in clinical tri-als.

A second issue is the continuing high cost of light sources. Using existingtechnologies it is not clear how this can be significantly reduced. Certainly,part of the cost comes from dealing with medical devices where significantinvestments are required to ensure safety and reliability. It is becoming clearthat, for some potentially important applications, it is the cost of the lightsource that is limiting, not the cost of the photosensitizer. Hence, there aresignificant opportunities for optical engineers and physicists to devise newdevices.

A further major limitation is that the specificity of photosensitizer target-ing of disease has not been high enough to use unlimited light safely. Attemptshave been made to improve this, for example, by linking the photosensitizerto antibodies to target tumors cells, but these have not been particularly suc-cessful to date. PDT beacons could change this significantly. However, it willbe challenging to get these into clinical trials and then through regulatoryapprovals, so that the preclinical results in vivo will have to be outstandingto make the effort worthwhile.

With respect to tissue response monitoring, there are many novel biopho-tonic imaging and spectroscopic techniques being developed for other appli-cations that could also be valuable in PDT: fluorescence imaging is a primeexample, while others include Raman spectroscopy (to report biochemicalchanges) and intravital second harmonic generation imaging (changes in tis-sue architecture). PDT monitoring will not drive the development of thesetechniques but will benefit from them.

Three other major aspects of PDT are worth mentioning:

– The parallel development of “photodynamic diagnostics,” particularly fortumor detection and localization [23]

– The combination of PDT with other therapeutic modalities, e.g., (fluo-rescence guided) tumor resection followed by PDT to eliminate residualminimal disease

– The extension of PDT into new applications, particularly to modify cellsor tissue function rather than destroying them: recent examples are formodifying bone growth and for targeting epilepsy.

In addition to these clinical applications, the principle of light-activateddrugs may also be useful for basic life sciences research. One example is the useof 2γ PDT to uncage growth factors in order to direct the growth of neuronsin tissue engineering [31].


17.7 Conclusions

Although PDT is over a century old, it is far from a mature science. Address-ing some of the challenges in PDT, particularly in the optical technologies anddosimetry, has been a major driving force in biophotonics as a whole, withsubstantial spin-off into other applications. For example, much of the earlywork in tissue optics was motivated by the need to understand and control thedistribution of light in PDT of solid tumors. In turn, this laid the foundationfor diffuse optical tomography and other therapeutic and diagnostic tech-niques. In the same way, studies of photosensitizer fluorescence led directly toimaging based on tissue autofluorescence: subsequently, attempts to improvethe image contrast stimulated work on targeted fluorophores, including mostrecently quantum dots.

It is reasonable to expect that PDT will continue, both academically andcommercially, especially with the extension into a diverse range of new ap-plications and using some of the new concepts outlined here. There is stillcertainly much to be done that will require the efforts of many different dis-ciplines.

Acknowledgments

The author thanks the following agencies for support of work illustratedhere: the National Cancer Institute of Canada (1O2 dosimetry and DOCTmonitoring), the National Institutes of Heath, US (prostate PDT and mPDT[(CA-43892)], beacons), the Canadian Institute for Photonic Innovations(2-photon PDT), and Photonics Research Ontario (PDT technology devel-opment). Also, thanks to many colleagues and students in the PDT programat the Ontario Cancer Institute and to our clinical and industry collaborators.

References

1. T. Patrice (ed.), Photodynamic Therapy, (Royal Society of Chemistry, UK,2003)

2. R. Ackroyd et al., Photochem. Photobiol. 74, 656 (2001)3. B.C. Wilson, S.G. Bown, in Handbook of Lasers and Applications, ed. by C.

Webb, J. Jones (IOP Publ, UK, 2004), p. 20194. T.J. Dougherty, J. Clin. Laser Med. Surg. 20, 3 (2002)5. T.N. Demidova, M.R. Hamblin, Int. J. Immunopathol. Pharmacol. 17, 245

(2004)6. B.W. Henderson, T.J. Dougherty, Photochem. Photobiol. 55, 145 (1992)7. N.L. Oleinick, H.H. Evans, Radiat. Res. 150, S146 (1998)8. Q. Peng et al., Photochem. Photobiol. 65, 235 (1997)9. W. Stummer et al., Acta Neurochir. Suppl. 88, 9 (2003)

10. M.S. Patterson, B.C. Wilson, in Modern Technology of Radiation Oncology, ed.by J. Van Dyke (Medical Physics Publishing, Madison, WI, USA, 1999), p. 941

334 B.C. Wilson

11. L. Brancaleon, H. Moseley, Lasers Med. Sci. 17, 173 (2002)12. K.S. Samkoe, D.T. Cramb, J. Biomed. Opt. 8, 410 (2003)13. A. Karotki et al., Photochem. Photobiol. 82, 443 (2006)14. Y. Mir et al., Photochem. Photobiol. Sci. 5, 1024 (2006)15. S.K. Bisland et al., Photochem. Photobiol. 80, 2 (2004)16. U. Mahmood, R. Weissleder, Mol. Cancer Ther. 2, 489 (2003)17. J. Chen et al., J. Am. Chem. Soc. 126, 11450 (2004)18. M. Ferrari, Nat. Rev. Cancer 5, 161 (2005)19. B.C. Wilson, in Photon-Based NanoScience and Technology, ed. by J.

Dubowski, S. Tanev (Springer, Netherlands, 2006), p. 12120. B.C. Wilson et al., Lasers Med. Sci. 12, 182 (1997)21. R.A. Weersink et al., J. Photochem. Photobiol. 79, 211 (2005)22. B.W. Pogue, G. Burke, Appl. Opt. 37, 7429 (1998)23. B.C. Wilson, in Handbook of Lasers and Applications, ed. by C. Webb, J. Jones

(IOP, UK, 2004), p. 208724. J.S. Dysart, M.S. Patterson, Photochem. Photobiol. Sci. 5, 73 (2006)25. M. Niedre et al., Photochem. Photobiol. 75, 382 (2002); 81, 941 (2005)26. M. Niedre et al., Cancer Res. 63, 7986 (2003)27. M. Niedre et al., Br. J. Cancer 92, 298 (2005)28. E. Moriyama et al., Photochem. Photobiol. 80, 242 (2004)29. H. Li et al., Lasers Surg. Med. 38, 754 (2006)30. V.X. Yang et al., Opt. Lett. 30, 1791 (2005)31. Y. Luo, M.S. Stoichet, Nat. Mater. 3, 249 (2004)

Index

1D photonic crystal, 1032D photonic crystal, 103

Active pixel sensor, 239Avidin, 200

Biosensor, 107, 199BSA, 120

Carotenoids, 3, 31, 42CARS:Coherent anti-Stokes Raman, 48CCD, 239Chlorophylls, 3, 29Chloroplasts, 2

DHM, 164DHM Life Cell Imaging, 171DHM numerical focus, 170DHM resolution, 170Digital Holographic Microscopy, 164Digital holographic reconstruction, 166Distal endoscopic ESPI, 158DNA, 114Double exposure subtraction ESPI, 152

Effective dielectric constant, 110Electronic Speckle Pattern Interferome-

try, 152ESPI, 152

Fluorescence, 49Fluorescent dye molecule, 87Functionalization, 111

Genetic engineering, 202

Genome, 177Genomics, 219Gram bacteria, 114Greenhouse effect, 15

Hydrogenases, 19

IgG, 117

Keratinocytes, 301

Label-free, 88Label-free optical biosensing, 109Laser tissue welding, 275Light cone, 104Light harvesting complexes, 6

Mach–Zehnder interferometer, 128Michelson-type interferometer, 128Microcavity, 94, 105, 110, 111, 121Microscopic Speckle Interferometry, 161Multi-photon microscopy, 47Mutant Type protein, 195

NDRM, 166Non diffractive reconstruction, 166

Ophthalmology, 127Optical tweezers, 249Optrodes, 204, 207

Phosphorescennce, 49Photonic bandgap, 105Photonic crystals, 101Photosynthesis, 1, 17Plasmid, 194

336 Index

Porous silicon, 107Proteomeics, 219Proteomics, 221Proximal endoscopic ESPI, 156

Quantum dots, 178

Second harmonic generation, 48Sensitivity vector, 153Setups for DHM, 165Silanization, 114Single photon avalanche diode, SPAD,

239

Singlet oxygen, 30Spatial phase shifting, 153SPS, 153Stable transfection, 194Streptavidin, 115

Thylakoids, 2, 6Transfection, 194Transient transfection, 194Two-photon cross-section, 58

Vector, 194

biological and medical physics, · biological and medical physics, ... they lie at the crossroads...

Documents